Synthesis, Optimization, Antifungal Activity, Selectivity, and CYP51 Binding of New 2-Aryl-3-azolyl-1-indolyl-propan-2-ols

A series of 2-aryl-3-azolyl-1-indolyl-propan-2-ols was designed as new analogs of fluconazole (FLC) by replacing one of its two triazole moieties by an indole scaffold. Two different chemical approaches were then developed. The first one, in seven steps, involved the synthesis of the key intermediate 1-(1H-benzotriazol-1-yl)methyl-1H-indole and the final opening of oxiranes by imidazole or 1H-1,2,4-triazole. The second route allowed access to the target compounds in only three steps, this time with the ring opening by indole and analogs. Twenty azole derivatives were tested against Candida albicans and other Candida species. The enantiomers of the best anti-Candida compound, 2-(2,4-dichlorophenyl)-3-(1H-indol-1-yl)-1-(1H-1,2,4-triazol-1-yl)-propan-2-ol (8g), were analyzed by X-ray diffraction to determine their absolute configuration. The (−)-8g enantiomer (Minimum inhibitory concentration (MIC) = IC80 = 0.000256 µg/mL on C. albicans CA98001) was found with the S-absolute configuration. In contrast the (+)-8g enantiomer was found with the R-absolute configuration (MIC = 0.023 µg/mL on C. albicans CA98001). By comparison, the MIC value for FLC was determined as 0.020 µg/mL for the same clinical isolate. Additionally, molecular docking calculations and molecular dynamics simulations were carried out using a crystal structure of Candida albicans lanosterol 14α-demethylase (CaCYP51). The (−)-(S)-8g enantiomer aligned with the positioning of posaconazole within both the heme and access channel binding sites, which was consistent with its biological results. All target compounds have been also studied against human fetal lung fibroblast (MRC-5) cells. Finally, the selectivity of four compounds on a panel of human P450-dependent enzymes (CYP19, CYP17, CYP26A1, CYP11B1, and CYP11B2) was investigated.


Introduction
Invasive fungal infections are related to a high mortality rate despite the availability of several antifungal drugs. Fungi species that belong to one of these four genera (Cryptococcus, Candida, Aspergillus, and Pneumocystis) kill about one and a half million people every year [1,2]. These infections have increased in frequency and severity over the last two decades as a result of an increasing number of immunocompromised hosts due to cancer chemotherapy, organ and bone marrow transplantation, human immunodeficiency virus, or therapy against autoimmune and inflammatory diseases especially with TNF inhibitors [3]. Candida albicans (C. albicans) is the major opportunistic pathogen of fungal infections and Candida spp. are the fourth most common nosocomial bloodstream pathogens in the USA with a mortality rate of 40% [1,2].
For many years, the polyene amphotericin B has been the only available antifungal drug for invasive fungal infections despite nephrotoxicity, infusion-related reactions, and other adverse effects [4]. To this day, approved antifungal agents belong to only four drug classes: Polyenes, azoles, echinocandins, and anti-metabolites [5]. Introduction of triazoles (e.g., fluconazole (FLC), itraconazole) offered new treatment options with excellent antifungal activity and a good safety profile. Prophylactic use of FLC has been developed to reduce the risk of Candida colonization and infection in high-risk patients with minimal toxicity [6,7]. Nevertheless, the emergence of C. krusei and C. glabrata infections in patients with bone marrow transplantation or neutropenia receiving FLC prophylaxis has been reported [8,9]. Development of Candida spp. resistance has also commonly been observed in HIV-infected patients who received prolonged FLC treatment for oropharyngeal candidiasis [2,10,11].
Fourth generation azoles (e.g., voriconazole, posaconazole, ravuconazole, isavuconazole) and echinocandins (e.g., caspofungin, micafungin, anidulafungin) approved for use or under clinical trials have proven to be less toxic than conventional amphotericin B, resulting in a significant advance in treatment of serious fungal infections. However, voriconazole interferes with many drugs and its prolonged use may expose severely immunocompromised patients to the risk of zygomycosis [12], while echinocandins are not available orally [13]. Consequently, most patients requiring therapy for invasive candidiasis are still placed on the safe, inexpensive, and orally available fluconazole. Despite recent advances in antifungal chemotherapy, the mortality rates of invasive fungal infections have remained unchanged and there is a necessity for the development of new antifungal compounds. Aiming at new and effective antifungal drugs, our attention was focused on the azole derivatives because of their low toxicity, broad spectrum of activity, and favorable pharmacokinetic properties [10,11,14].
In order to reduce the number of reaction steps, we investigated another synthetic pathway called route 2 (Scheme 2). First, imidazole and 1H-1,2,4-triazole were alkylated by commercially available halogenoacetophenones using K2CO3 in CH3CN under microwave heating (85 °C, 50 W) to give the target ketones 9a-f, which were converted to oxiranes 10a-f in the presence of TMSOI and an aqueous solution of sodium hydroxide in toluene under microwave heating (80 °C, 50 W). The reaction times of these two steps (alkylation and epoxidation) are greatly reduced by using microwave heating as we have already reported [27]. Finally, the ring opening of 10a-f with indole using NaH in DMSO gave the derivatives 8g and 11a-e. A similar ring opening of the intermediates 9d and 9e by indoles with methyl groups in position 2 and/or 3 furnished the target derivatives 11fk. In the case of the synthesis of compound 8g, route 2 gave a great improvement of the overall yield (route 1: 5% vs. route 2: 29%).  In order to reduce the number of reaction steps, we investigated another synthetic pathway called route 2 (Scheme 2). First, imidazole and 1H-1,2,4-triazole were alkylated by commercially available halogenoacetophenones using K 2 CO 3 in CH 3 CN under microwave heating (85 • C, 50 W) to give the target ketones 9a-f, which were converted to oxiranes 10a-f in the presence of TMSOI and an aqueous solution of sodium hydroxide in toluene under microwave heating (80 • C, 50 W). The reaction times of Pharmaceuticals 2020, 13, 186 5 of 32 these two steps (alkylation and epoxidation) are greatly reduced by using microwave heating as we have already reported [27]. Finally, the ring opening of 10a-f with indole using NaH in DMSO gave the derivatives 8g and 11a-e. A similar ring opening of the intermediates 9d and 9e by indoles with methyl groups in position 2 and/or 3 furnished the target derivatives 11f-k. In the case of the synthesis of compound 8g, route 2 gave a great improvement of the overall yield (route 1: 5% vs. route 2: 29%).

X-ray Structural Studies
In order to confirm the absolute configuration of the synthesized compounds 8g, X-ray structure analyses for the two derivatives were performed. The atom labeling and thermal ellipsoids of (+)-8g and (−)-8g are shown in Figures 2 and 3. Two independent molecules, designated as A and B, were found in the asymmetric crystallographic unit of (+)-8g and (−)-8g. The configuration of (+)-8g and (−)-8g was determined by observing and calculating the F(+)/F(−) ratios of Bijvoet pairs with the mean F value of each independent reflection. Based on the results, the absolute configuration at C-9 and C-59 in (+)-8g was determined to be R. The (−)-8g enantiomer was found with the S-absolute configuration. Figure 2. View of the crystal structure of (R)-8g with our numbering scheme, displacement ellipsoids are drawn at the 30% probability level.

X-ray Structural Studies
In order to confirm the absolute configuration of the synthesized compounds 8g, X-ray structure analyses for the two derivatives were performed. The atom labeling and thermal ellipsoids of (+)-8g and (−)-8g are shown in Figures 2 and 3. Two independent molecules, designated as A and B, were found in the asymmetric crystallographic unit of (+)-8g and (−)-8g. The configuration of (+)-8g and (−)-8g was determined by observing and calculating the F(+)/F(−) ratios of Bijvoet pairs with the mean F value of each independent reflection. Based on the results, the absolute configuration at C-9 and C-59 in (+)-8g was determined to be R. The (−)-8g enantiomer was found with the S-absolute configuration.

X-ray Structural Studies
In order to confirm the absolute configuration of the synthesized compounds 8g, X-ray structure analyses for the two derivatives were performed. The atom labeling and thermal ellipsoids of (+)-8g and (−)-8g are shown in Figures 2 and 3. Two independent molecules, designated as A and B, were found in the asymmetric crystallographic unit of (+)-8g and (−)-8g. The configuration of (+)-8g and (−)-8g was determined by observing and calculating the F(+)/F(−) ratios of Bijvoet pairs with the mean F value of each independent reflection. Based on the results, the absolute configuration at C-9 and C-59 in (+)-8g was determined to be R. The (−)-8g enantiomer was found with the S-absolute configuration.  View of the crystal structure of (S)-8g with our numbering scheme, displacement ellipsoids are drawn at the 30% probability level.

In Vitro Anti-Candida Activity
All anti-Candida activities of propanol derivatives 8a-g and 11a-m are presented in Table 1. Concerning the expression of the results during the evaluation of the compounds, we deliberately chose to calculate a more selective minimal inhibitory concentration (MIC = IC80, µ g/mL) than an IC50 in order to highlight some very active compounds. Concerning the 1H-1,2,4-triazole sub-series, substitutions of the phenyl ring by halogens greatly impacted anti-C. albicans activities. The 4trifluoromethylated 8f, 4-chlorinated 11b, 4-brominated 11c, and 2,4-difluorinated 11d compounds had activities comparable with FLC (MIC = 0.020 µ g/mL on C. albicans CA98001), while the 4fluorinated derivative 11a was less active (MIC = 0.210 µ g/mL). The most favorable substitution corresponded to the 2,4-dichlorinated phenyl ring for compound 8g. The asymmetric carbon of the propanol chain also played a very important role in the activity against C. albicans. Indeed, activity of the (+)-(R)-enantiomer was similar to FLC (MIC = 0.023 µ g/mL), whereas the (−)-(S)-enantiomer had an action comparable with the racemic mixture (MIC values of 0.000256 and 0.000259 µ g/mL, respectively).
In the imidazole sub-series, there was little difference in activity depending on the substitution of the phenyl ring because all compounds had activities similar to that of FLC except for the 2,4difluorinated derivative 8d, which was 57 times more active (MIC = 0.00035 µ g/mL). There was no correlation between the imidazole and 1H-1,2,4-triazole series according to the substitution of the phenyl ring. Compounds of 1H-1,2,3-triazole and 2H-1,2,3-triazole series with 2,4-dichlophenyl substitution 11l-m were totally inactive. This pharmacomodulation of the azolyl unit highlighted the necessity of the 1H-1,2,4-triazole or imidazole moiety to promote activity against C. albicans. The inactivity of 1H-1,2,3-triazole and 2H-1,2,3-triazole derivatives could be explained by the impossibility of these rings to coordinate with the iron atom of CYP51.
The last work of pharmacomodulation focused on the presence of the methyl group in position 2 and/or 3 of the indole ring. With a 2,4-difluorinated phenyl group, the presence of a methyl at position 2 on the indole ring did not change the activity (11f, MIC = 0.022 µ g/mL); on the other hand, the introduction of a methyl at position 3 enhanced the activity level (11h, MIC = 0.011 µ g/mL). In the case of two methyl groups at positions 2 and 3, the activity was reduced (11j, MIC = 0.157 µ g/mL). The activities of 2,4-dichloro derivatives were generally better than those encountered with 2,4difluorinated products. However, none of the proposed pharmacomodulations retained the activity Figure 3. View of the crystal structure of (S)-8g with our numbering scheme, displacement ellipsoids are drawn at the 30% probability level.

In Vitro Anti-Candida Activity
All anti-Candida activities of propanol derivatives 8a-g and 11a-m are presented in Table 1. Concerning the expression of the results during the evaluation of the compounds, we deliberately chose to calculate a more selective minimal inhibitory concentration (MIC = IC 80 , µg/mL) than an IC 50 in order to highlight some very active compounds. Concerning the 1H-1,2,4-triazole sub-series, substitutions of the phenyl ring by halogens greatly impacted anti-C. albicans activities. The 4-trifluoromethylated 8f, 4-chlorinated 11b, 4-brominated 11c, and 2,4-difluorinated 11d compounds had activities comparable with FLC (MIC = 0.020 µg/mL on C. albicans CA98001), while the 4-fluorinated derivative 11a was less active (MIC = 0.210 µg/mL). The most favorable substitution corresponded to the 2,4-dichlorinated phenyl ring for compound 8g. The asymmetric carbon of the propanol chain also played a very important role in the activity against C. albicans. Indeed, activity of the (+)-(R)-enantiomer was similar to FLC (MIC = 0.023 µg/mL), whereas the (−)-(S)-enantiomer had an action comparable with the racemic mixture (MIC values of 0.000256 and 0.000259 µg/mL, respectively).
In the imidazole sub-series, there was little difference in activity depending on the substitution of the phenyl ring because all compounds had activities similar to that of FLC except for the 2,4-difluorinated derivative 8d, which was 57 times more active (MIC = 0.00035 µg/mL). There was no correlation between the imidazole and 1H-1,2,4-triazole series according to the substitution of the phenyl ring. Compounds of 1H-1,2,3-triazole and 2H-1,2,3-triazole series with 2,4-dichlophenyl substitution 11l-m were totally inactive. This pharmacomodulation of the azolyl unit highlighted the necessity of the 1H-1,2,4-triazole or imidazole moiety to promote activity against C. albicans. The inactivity of 1H-1,2,3-triazole and 2H-1,2,3-triazole derivatives could be explained by the impossibility of these rings to coordinate with the iron atom of CYP51. Table 1. Anti-Candida activity of propanol derivatives 8a-g and 11a-m.
indole ring may have been the cause for this decrease in activity, by imposing an unfavorable position of the indole in the active site of the enzyme.
Overall, it emerged that the synthesized compounds showed activity against C. albicans comparable with or greater than fluconazole. For example, difluoro derivatives (8d and 11h) and dichloro derivatives (8g and 11i) were the most active derivatives. In the particular case of compound 8g, (±)-8g and its (−)-(S) enantiomer were ≈80 times more active than FLC against C. albicans CA98001. The activity spectrum on Candida species (Table 2) shows that compound 8g and more particularly its (S)-enantiomer maintained a better activity than FLC against the various strains tested, especially on those naturally resistant to FLC. For example, on C. krusei, the MIC values of (S)-8g were between 0.167 and 0.039 µg/mL. The last work of pharmacomodulation focused on the presence of the methyl group in position 2 and/or 3 of the indole ring. With a 2,4-difluorinated phenyl group, the presence of a methyl at position 2 on the indole ring did not change the activity (11f, MIC = 0.022 µg/mL); on the other hand, the introduction of a methyl at position 3 enhanced the activity level (11h, MIC = 0.011 µg/mL). In the case of two methyl groups at positions 2 and 3, the activity was reduced (11j, MIC = 0.157 µg/mL). The activities of 2,4-dichloro derivatives were generally better than those encountered with 2,4-difluorinated products. However, none of the proposed pharmacomodulations retained the activity obtained with the unsubstituted indole. In this case, the inactivity of the 2,3-dimethylindole unit was even more pronounced. The conjugated steric effect of the methyl groups on positions 2 and 3 on the indole ring may have been the cause for this decrease in activity, by imposing an unfavorable position of the indole in the active site of the enzyme.
Overall, it emerged that the synthesized compounds showed activity against C. albicans comparable with or greater than fluconazole. For example, difluoro derivatives (8d and 11h) and dichloro derivatives (8g and 11i) were the most active derivatives. In the particular case of compound 8g, (±)-8g and its (−)-(S) enantiomer were ≈80 times more active than FLC against C. albicans CA98001.
The activity spectrum on Candida species (Table 2) shows that compound 8g and more particularly its (S)-enantiomer maintained a better activity than FLC against the various strains tested, especially on those naturally resistant to FLC. For example, on C. krusei, the MIC values of (S)-8g were between 0.167 and 0.039 µg/mL.
While the presence of a methyl group on the indole ring globally increased the cytotoxicity of the 2,4-difluoro derivatives 11h and 11j, the introduction of a methyl group in position 2 into 2,4-dichlorinated products seemed to give less cytotoxic compounds (e.g. 11g, 11k).
The enantiomers of 8g were evaluated for inhibitory activity against retinoic acid 4-hydroxylase (CYP26A1) using a cellular assay. The standards for comparison in the CYP26A1 assay were the broad spectrum CYP inhibitor KTC and liarozole. The (+)-(R)-8g was a very weak inhibitor of CYP26A1 (IC 50  Racemic compound 8g and its enantiomers were also tested for their potential inhibitory effect against CYP11B1 and CYP11B2. 8g showed no inhibitory activity against CYP11B1 and only a slight inhibitory effect against CYP11B2 (at high concentrations). Both enantiomers showed no inhibitory activity against CYP11B1. As mentioned before [37], KTC was a strong inhibitor of CYP11B1; on the other hand, FLC was a minor inhibitor. Only (+)-(R)-8g showed a slight inhibitory effect against CYP11B2 (at high concentrations) while KTC exerted 100% inhibition in another similar study [38].

Molecular Docking
The predicted binding pose of (S)-8g ( Figure 4) was to some degree consistent with the binding conformation of co-crystallized posaconazole in the Protein Data Bank (PDB) structure (PDB ID: 5FSA [39]) with the nitrogen atom in the triazole ring interacting with the Fe atom in the heme group. However, the predicted binding pose of (R)-8g ( Figure 5) was not consistent with the posaconazole pose in its co-crystal structure where the triazole ring was more parallel with the heme group, rather than perpendicular to the heme, that is, a less favorable angle and subsequently a weaker binding interaction would be expected. Therefore, the (S)-8g generated a better docking score and more credible binding pose compared with (R)-8g, which is consistent with the biological experiments described in 2.3.1. Pharmaceuticals 2020, 13, x FOR PEER REVIEW 12 of 32

Molecular Dynamics Simulation
In order to confirm the first results obtained by molecular docking, a molecular dynamics (MD) approach was also applied to CaCYP51-azole complexes.
Both enantiomers formed primarily hydrophobic interactions including Tyr118, Phe126, and Tyr132 for both enantiomers, Ile304 for (R)-8g, and Leu376 for (S)-8g (Figures 6 and 7). In both enantiomers the dichlorophenyl ring forms a π-π stacking interaction, with Phe126 and Tyr132 for the (R)-8g and (S)-8g enantiomer, respectively. Of note is the interaction of the triazole N with the

Molecular Dynamics Simulation
In order to confirm the first results obtained by molecular docking, a molecular dynamics (MD) approach was also applied to CaCYP51-azole complexes.
Both enantiomers formed primarily hydrophobic interactions including Tyr118, Phe126, and Tyr132 for both enantiomers, Ile304 for (R)-8g, and Leu376 for (S)-8g (Figures 6 and 7). In both enantiomers the dichlorophenyl ring forms a π-π stacking interaction, with Phe126 and Tyr132 for the (R)-8g and (S)-8g enantiomer, respectively. Of note is the interaction of the triazole N with the

Molecular Dynamics Simulation
In order to confirm the first results obtained by molecular docking, a molecular dynamics (MD) approach was also applied to CaCYP51-azole complexes.
Both enantiomers formed primarily hydrophobic interactions including Tyr118, Phe126, and Tyr132 for both enantiomers, Ile304 for (R)-8g, and Leu376 for (S)-8g (Figures 6 and 7). In both enantiomers the dichlorophenyl ring forms a π-π stacking interaction, with Phe126 and Tyr132 for the (R)-8g and (S)-8g enantiomer, respectively. Of note is the interaction of the triazole N with the heme iron, which occurred for 75% of the simulation time for (R)-8g and 100% of the simulation time for (S)-8g, suggesting improved binding of the (S)-enantiomer with the heme active site.
Pharmaceuticals 2020, 13, x FOR PEER REVIEW 13 of 32 heme iron, which occurred for 75% of the simulation time for (R)-8g and 100% of the simulation time for (S)-8g, suggesting improved binding of the (S)-enantiomer with the heme active site.
(R)-8g (S)-8g The reason for the reduced binding interaction of the (R)-enantiomer of 8g can be seen from the 3D image after MD simulation (Figure 8). For optimal binding of the triazole nitrogen with the heme iron, the triazole should have been positioned perpendicular to the heme as observed for the (S)-8g enantiomer; however, the triazole of the (R)-8g enantiomer was positioned parallel to the heme, reducing optimal binding. The unfavorable position of the (R)-enantiomer triazole also increased the distance between the triazole N and the heme iron to 3.11 Å compared with 2.43 Å for the optimally positioned (S)-enantiomer. The reason for the reduced binding interaction of the (R)-enantiomer of 8g can be seen from the 3D image after MD simulation (Figure 8). For optimal binding of the triazole nitrogen with the heme iron, the triazole should have been positioned perpendicular to the heme as observed for the (S)-8g enantiomer; however, the triazole of the (R)-8g enantiomer was positioned parallel to the heme, reducing optimal binding. The unfavorable position of the (R)-enantiomer triazole also increased the distance between the triazole N and the heme iron to 3.11 Å compared with 2.43 Å for the optimally positioned (S)-enantiomer. The reason for the reduced binding interaction of the (R)-enantiomer of 8g can be seen from the 3D image after MD simulation (Figure 8). For optimal binding of the triazole nitrogen with the heme iron, the triazole should have been positioned perpendicular to the heme as observed for the (S)-8g enantiomer; however, the triazole of the (R)-8g enantiomer was positioned parallel to the heme, reducing optimal binding. The unfavorable position of the (R)-enantiomer triazole also increased the distance between the triazole N and the heme iron to 3.11 Å compared with 2.43 Å for the optimally positioned (S)-enantiomer. The overlap of both enantiomers of 8g and posaconazole shows that the (S)-enantiomer aligned with the positioning of posaconazole within both the heme and access channel binding sites, whereas for the (R)-enantiomer the dichlorophenyl was outside the small aryl binding pocket and the indole was directed away from the access channel ( Figure 9).  The overlap of both enantiomers of 8g and posaconazole shows that the (S)-enantiomer aligned with the positioning of posaconazole within both the heme and access channel binding sites, whereas for the (R)-enantiomer the dichlorophenyl was outside the small aryl binding pocket and the indole was directed away from the access channel ( Figure 9). The overlap of both enantiomers of 8g and posaconazole shows that the (S)-enantiomer aligned with the positioning of posaconazole within both the heme and access channel binding sites, whereas for the (R)-enantiomer the dichlorophenyl was outside the small aryl binding pocket and the indole was directed away from the access channel ( Figure 9).

Chemistry
The microwave reactor was a monomode system (Discover, CEM) with focused waves operating at 2.45 GHz. Melting points were determined on an Electrothermal IA9300 melting point digital apparatus and were reported uncorrected. Infrared (IR) spectra were obtained in KBr pellets or neat liquid films with a Perkin-Elmer Paragon FTIR 1000 PC spectrometer. 1 H and 13 C-NMR spectra were recorded on a Bruker AC 250 or AVANCE 400 spectrometer in DMSO-d 6 as solvent. Chemical shifts (δ) were reported in part per million (ppm) relative to tetramethylsilane as internal standard. The splitting pattern abbreviations are as follows: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet; dd, doublet of doublet; td, triplet of doublet. Coupling constants J were given in Hz. Mass spectral (MS) analysis was performed on a quadrupole H.P. 5889A instrument using EI mode at 70 eV. Microanalyses were performed on a Perkin-Elmer CHN 240 apparatus. All reactions were monitored by TLC, using 0.25 mm-thick precoated silica gel plates (E. Merck, Darmstadt, Germany). Compounds were purified by column chromatography using silica gel 60 as a stationary phase. All common chemicals and solvents utilized were reagent grade and purchased from Sigma-Aldrich (Saint Quentin, France).
The synthesis of key intermediates 9a-e and 10a-e was previously described by IICiMed (Nantes) [27].

General Procedure for the Synthesis of Benzotriazole Derivatives 5a-f
1-(1H-Benzotriazol-1-yl-methyl)-1H-indole (4) (1.88 g, 7.57 mmol) in THF (50 mL) was cooled to -78 • C under argon, and n-BuLi (1.6 M in THF, 5.67 mL, 9.1 mmol) was added dropwise, then the mixture was stirred at -78 • C for 1 h. At the end of this period, the corresponding ethyl benzoate (9.46 mmol) was added dropwise and the mixture was slowly warmed to rt over a period of 12 h. Then saturated aqueous ammonium chloride solution (35 mL) was added, the mixture was diluted with H 2 O (30 mL), and extracted with diethyl ether (3 × 40 mL). The organic layer was washed with brine (40 mL), dried over anhydrous sodium sulfate, and the volatile fraction was evaporated under reduced pressure. The residue was purified by silica gel column chromatography (dichloromethane).

General Procedure for the Synthesis of Oxiranes 7a-f
Trimethylsulfoxonium iodide (0.44 g, 2.02 mmol) and sodium hydroxide (6 g, 150 mmol, 48% in aqueous solution) were added to a solution of the corresponding N-substituted indole 6a-f (1.41 mmol) in dichloromethane (5 mL). Then the reaction mixture was stirred and refluxed for 48 h. At the end of this period, the mixture was diluted with H 2 O (10 mL) and extracted with dichloromethane (3 × 10 mL). The organic layer was washed with H 2 O (15 mL), dried over anhydrous sodium sulfate, and the volatile fraction was evaporated under reduced pressure. The residue was purified by silica gel column chromatography (dichloromethane/hexane 1:1, v/v).

General Procedure for the Synthesis of Imidazole Derivatives 8a-e
Potassium carbonate (0.25 g, 1.83 mmol) and 1H-imidazole (0.12 g, 1.82 mmol) were added to a solution of the corresponding oxirane 7a-e (0.64 mmol) in dimethylformamide (20 mL). Then the reaction mixture was stirred and refluxed for 7 h. At the end of this period, the mixture was diluted with H 2 O (20 mL) and extracted with EtOAc (3 × 40 mL). The organic layer was washed with brine (40 mL), dried over anhydrous sodium sulfate, and the volatile fraction was evaporated under reduced pressure. The residue was purified by silica gel column chromatography (dichloromethane/ethanol 19:1, v/v).

General Procedure for the Sysnthesis of Triazole Derivatives 11l and 11m
A procedure similar to 11a was used to prepare compounds 11l and 11m.

Anti-Candida In Vitro Assay
Test fungal strains were obtained from the American Type Culture Collection (ATCC) or were clinical isolates from the Laboratory of Parasitology and Medical Mycology, Centre Hospitalier Universitaire (CHU) of Nantes. The strains were maintained on Sabouraud agar slants and were subcultured 24 h before used. The activity of the target compounds 8a-g and 11a-m was determined by the method previously described [48]. Briefly, molecules were diluted in RPMI 1640 medium supplemented with 0.165 M morpholinopropanesulphonic acid (Sigma-Aldrich, Saint-Quentin Fallavier, France), 2% glucose, and antibiotics. Candida suspensions were prepared in RPMI 1640 medium (Sigma-Aldrich, Saint-Quentin Fallavier, France) adjusted to give a final concentration of 10 3 cells/mL, and a 96-well microplate (Nunc, D. Dutscher, Brumath, France) was seeded with 100 µL. Each concentration of molecules (100 µL) to be tested was added (in triplicate) and plates were incubated at 37 • C for 24 h. The cellular viability was evaluated on the Fluorolite 1000 (Dynatech, France) with an excitation at 550 nm and an emission at 590 nm after a 4 h incubation with 10 µL of Alamar Blue ® . The minimal inhibitory concentration (MIC) is the concentration that inhibited 80% of the cell growth and was determined by linear regression analysis. MIC was expressed as the mean of the triplicate values. KTC and FLC were used as standards.

In Vitro Inhibition of Aromatase
The assay was performed according to previously described methods [49][50][51] by monitoring the enzyme activity by measuring the 3 H 2 O formed from [1β-3 H]androstenedione during aromatization. In brief, the reaction mixture, containing [1β-3 H]androstenedione (0.08 µCi, 15 nM), unlabeled androstenedione (485 nM), the NADPH-generating system (2 mM), the inhibitor (0-100 µM), and phosphate buffer (0.05 M, pH 7.4), was preincubated for 5 min at 30 • C. Microsomal protein (0.1 mg) was added to start the reaction. After incubation of 15 min at 30 • C, the reaction was stopped by adding 200 µL of a cold HgCl 2 solution (1 mM). After the addition of 200 µL of an aqueous dextran-coated charcoal (DCC) suspension (2%), the vials were shaken then centrifuged to separate the charcoal-absorbed steroids. Aliquots of the supernatant were assayed for 3 H 2 O by counting in a scintillation mixture using a LKB-Wallac β-counter.

In Vitro Inhibition of 17α-Hydroxylase/17,20-Lyase
The assay was performed similar to described methods [29,52]. In summary, the reaction mixture containing progesterone (1.25 mM), NADPH (125 nmol), the inhibitor, and phosphate buffer (pH 7.4) was preincubated for 5 min at 32 • C. Microsomal protein was added to start the reaction. at different concentrations, and equal volumes of ethanolic solutions were used in all cases (including controls). Final concentrations of inhibitors ranged from 100 nM to 5 µM. Cells were preincubated with the respective inhibitor solutions for 15 min prior to the addition of steroid substrates (100 nM DOC or 100 nM RSS, respectively). For detection of steroids, 1% of the total substrate amount was radioactively labeled as [ 3 H]RSS, [ 3 H]DOC, or [ 14 C]DOC. After 4 h (MB164) or 24 h (SZ1) cultures were extracted with chloroform and dried under vacuum. The residues were dissolved in 10 µL chloroform and spotted onto glass-backed silica-coated HPTLC plates (Kieselgel 60 F 254 , Merck; Darmstadt, Germany). In addition, small amounts of non-radioactive steroids were spotted as references. The HPTLC was developed twice in chloroform/methanol/water (300:20:1, v/v/v) and steroids identified after exposure to Fuji imaging plates and quantitated on a phosphoimager (BAS-2500, Fuji; Stamford, CT, USA).

Molecular Docking
First molecular docking was performed for the compound (R)-8g and (S)-8g to assess their binding feasibility and difference. A six-step procedure was applied. (1) Three crystal structures of C. albicans CYP51 (PDB code: 5TZ1, 5FSA, 5V5Z) were taken from Pocketome [55], in which different PDB structures of the same protein were superimposed by the binding sites to take the induce-fit effect into consideration. (2) All structures were combined to calculate grid potential ensembles for the docking of (R)-8g and (S)-8g. The grid maps were calculated on a 0.5 Å 3D grid, containing: (i) van der Waals potentials; (ii) electrostatic potentials; (iii) hydrogen bonding potentials; and (iv) hydrophobic potential grids. (3) The structures of ligands were taken from crystallography [43]. (4) Four distinctive starting poses were generated for sampling. (5) The ligand was then sampled in the pre-generated grid potential ensembles through biased probability Monte Carlo sampling method [56] to optimize the positional variables of the ligand. (6) After the sampling, the ligand conformation was re-scored with the full-atom ICM scoring function [57]. All docking and scoring were performed in ICM-Pro v3.8-7c.

Molecular Dynamics Simulation
Molecular dynamics simulation was performed as previously described [21]. Briefly, docking studies were performed using MOE [40] and CaCYP51 (PDB 5FSA [39]) to generate pdb files of the CaCYP51 crystal structure complexed with (R)-8g and (S)-8g. Molecular dynamics simulations were run on the CaCYP51-ligand complexes with the pdb files first optimized with protein preparation wizard in Maestro by assigning bond orders, adding hydrogen, and correcting incorrect bond types. A default quick relaxation protocol was used to minimise the MD systems with the Desmond program [41]. Force-field parameters for the complexes were assigned using the OPLS_ 2005 forcefield, that is, a 100 ns molecular dynamic run in the NPT ensemble (T = 300 K) at a constant pressure of 1 bar. Energy and trajectory atomic coordinate data were recorded at each 1.2 ns.

Conclusions
A series of 2-aryl-3-azolyl-1-indolyl-propan-2-ols was designed as new analogs of FLC by replacing one of its two triazole moieties by an indole scaffold. A first chemical approach was developed in seven steps, involving the synthesis of the key intermediate 1-(1H-benzotriazol-1-yl)methyl-1H-indole 4 and the final opening of oxiranes 7 by imidazole or 1H-1,2,4-triazole. A shorter process was also developed to access the target compounds in only three steps, this time with the ring opening by indole and analogs. Twenty azole derivatives were synthesized and tested against C. albicans and other Candida species. Thirteen compounds demonstrated a high level of activity against C. albicans CA98001, with MIC < 0.027 µg/mL, by comparison with the MIC value of FLC (MIC = 0.020 µg/mL). The in vitro cytotoxicity of compounds 8a-g and 11a-k was further evaluated on MRC-5 cells.
Overall, (S)-8g had a pharmacological profile to pursue further biological investigations such as a murine candidiasis assay and cytochrome P450 inhibition assays (e.g. CYP3A4, CYP2D6). Recently, we published extended biological exploration of 8g (in vitro and in vivo assays) and then we confirmed the full potential of this molecule [28]. Another major challenge is the emergence of resistance to azole antifungals among Candida species [58]. Then a multi-disciplinary approach with joint expertise and networks is crucial to assist efficiently medicinal chemists for new optimizations. According to the 3D structures of CaCYP51 and our results obtained (Figure 9), further pharmacomodulation works should be performed to explore both the position of the azolyl chain on indole scaffold and the introduction of a long chain to occupy the space of the access channel (as does the N-arylpiperazine chain of posaconazole). In this regard, NMR will be of special interest for studying ligand access channels in cytochrome P450 enzymes [59][60][61][62].
The last point is that the antifungal azoles targeting CaCYP51 could be a starting point for developing new treatments in some particular diseases such as human infections with protozoa (Trypanosomatidae) [63] and primary amoebic meningoencephalitis (PAM) [64]. The needs for new treatments are enormous in the field of parasitic diseases, and CYP51 is a promising target for further drug development.