Silver(I) complexes with voriconazole as promising anti-Candida agents

By Mia Stanković, Sanja Skaro Bogojevic, Jakob Kljun, Žiko Milanović, Nevena Lj. Stevanović, Jelena Lazic, Sandra Vojnovic, Iztok Turel, Miloš I. Djuran, Biljana Đ. Glišić

Abstract

Recognizing that metal ions play an important role in modifying the pharmacological properties of known organic-based drugs, the present manuscript addresses the complexation of the antifungal agent voriconazole (vcz) with the biologically relevant silver(I) ion as a strategy for the development of new antimycotics. The synthesized silver(I) complexes with vcz were characterized by mass spectrometry, IR, UV–Vis and NMR spectroscopy and single-crystal X-ray diffraction analysis. The crystallographic results showed that complexes {[Ag(vcz)(H2O)]CH3SO3}n (1), {[Ag(vcz)2]BF4}n (2) and {[Ag(vcz)2]PF6}n (3) have polymeric structures in the solid state, in which silver(I) ions have a distorted tetrahedral geometry. On the other hand, DFT calculations revealed that the investigated silver(I) complexes 1–3 in DMSO exist as linear [Ag(vcz-N2)(vcz-N19)]+ (1a), [Ag(vcz-N2)(vcz-N4)]+ (2a) and [Ag(vcz-N4)2]+ (3a) species, respectively. The evaluated complexes showed an enhanced anti-Candida activity compared to the parent drug with minimal inhibitory concentration (MIC) values in the range of 0.02–1.05 μM. In comparison with vcz, the corresponding silver(I) complexes showed better activity in prevention hyphae and biofilm formation of C. albicans, indicating that they could be considered as promising agents against Candida that significantly inhibit its virulence. Also, these complexes are much better inhibitors of ergosterol synthesis in the cell membrane of C. albicans at the concentration of 0.5 × MIC. This is also confirmed by a molecular docking, which revealed that complexes 1a – 3a showed better inhibitory activity than vcz against the sterol 14α-demethylase enzyme cytochrome P450 (CYP51B), which plays a crucial role in the formation of ergosterol.

Graphical abstract

Considering the more favorable activity of silver(I) complexes with voriconazole compared to this antifungal agent, such complexes should be further investigated as candidates for potential use in therapy against lethal Candida infections.

Keywords

Silver(I) complexes – Voriconazole – Coordination polymers – DFT calculations – Anti-Candida activity – CYP51B

1. Introduction

Fungal infections have a profound impact on global human health with incidences rising notably over the last few decades. About 1.5–2 million people die of a fungal infection each year and most of this mortality is caused by species belonging to four genera of fungi: Aspergillus, Candida, Cryptococcus, and Pneumocystis [[1], [2], [3]]. Moreover, it was recently reported that the incidence of fungal diseases is substantially more frequent than previously thought [4]. Currently, five major categories of antifungal drugs, namely polyenes, azoles, allylamines, caspofungins, and pyrimidine analogues are available to treat invasive fungal infections [5]. Azoles are the most used antifungal drugs, since they exhibit their activity by inhibiting the fungal lanosterol 14α-demethylase, an enzyme that catalyses a key step in the biosynthesis of ergosterol, a major component of fungal plasma membranes [6]. These antifungal agents contain either two or three nitrogen atoms in the azole ring and are classified as imidazoles or triazoles, respectively. While most imidazoles are formulated only for topical use, usually because of toxicity or limited bioavailability, the antifungal agents that are licensed for clinical use in invasive fungal diseases are all triazoles [7].
Voriconazole (vcz, Fig. 1), along with posaconazole, ravuconazole and albaconazole, represent a class of triazole agents that has been developed to address the rising incidence of invasive fungal infections and the emergence of fungal resistance [8]. This chiral drug is an antifungal azole with a wide spectrum of activity, often applied in the treatment of invasive fungal infections [4,9]. The main disadvantage of the use of voriconazole in therapy is the inter-individual variation of its plasma levels caused by different factors which include liver function, polymorphisms in cytochrome P450 isoenzymes [9,10], pharmacokinetics, and preconditions such as liver disease or cancer [9,[11], [12], [13]]. Considering that metal ions can modify the pharmacological properties of organic-based drugs [[14], [15], [16], [17]], most of which can be coordinated with a metal ion, it was assumed that the limited efficacy of voriconazole in the treatment of invasive fungal infections could be overcome by its complexation with silver(I) ion. This assumption is supported by the fact that voriconazole possesses, along with the triazole ring, a hydroxyl group and a pyrimidine ring as potential binding sites (Fig. 1). Considering this, it may show distinctive modes of coordination with metal ions and such versatility of coordination may lead to unusual architectures and properties of metal-voriconazole complexes [18]. In addition, it is worth mentioning that the coordination of different organic-based ligands with silver(I) ions can achieve a prolonged release of these ions at the site of infection exerting antibacterial, antifungal, and anticancer activities [[19], [20], [21], [22], [23]].

Fig. 1. Structural formula of voriconazole (vcz).

Many previous crystallographic studies of the silver(I) complexes have clearly shown that the the overall architecture of the crystal structure is not only controlled by the nature of the ligands present in the moiety, but by different counter anions of the silver(I) salts used as the starting material in the synthesis [[24], [25], [26]]. Moreover, it was found that counter anion of the AgX have an influence on the antimicrobial properties of the evaluated silver(I) complexes [21,25,27]. The influence of the nature of silver(I) salts, AgX (X = CF3SO3−, NO3−, ClO4− and SbF6−), on the coordination mode of silver(I) ion to voriconazole has been investigated and the antifungal activities of the resulted silver(I) coordination compounds and the parent vcz drug have been compared [18,28,29]. The following silver(I)-voriconazole complexes were previously reported: {[Ag(vcz)2]CF3SO3}n [18], {[Ag(vcz)2]ClO4}n [28], {[Ag(vcz)2]NO3}n [28] and {[Ag(vcz)2]SbF6}n [29]. In these complexes, silver(I) ion binds to voriconazole in a 1 : 2 metal : ligand ratio, whereby this antifungal drug exhibits different coordination modes. The silver(I) ion in {[Ag(vcz)2]CF3SO3}n complex is coordinated by two nitrogen atoms of pyrimidine and two nitrogen atoms of triazole from different voriconazole molecules to form a tetrahedral structure [18]. In {[Ag(vcz)2]ClO4}n complex, silver(I) ion exhibits a four coordinated distorted tetrahedral geometry. Two voriconazole molecules in this complex have different coordination modes: one is coordinated to silver(I) ion in a monodentate fashion through one triazole nitrogen atom and the other in a tridentate T-shaped fashion through two triazole nitrogen atoms and one pyrimidine nitrogen [28]. Complex {[Ag(vcz)2]NO3}n contains two crystallographically independent silver(I) ions bound by five nitrogen atoms in a distorted trigonal bipyramidal geometry and by four nitrogen atoms in a trigonal pyramidal geometry. In this complex, voriconazole showed mono-, di- and tridentate coordination modes [28]. Silver(I) ions in {[Ag(vcz)2]SbF6}n complex are coordinated by four nitrogen atoms and adopt the same distorted tetrahedral geometry [29], as was observed for silver(I)-voriconazole complexes containing triflate [18] and perchlorate [28] counter anions.
The biological evaluation of silver(I)-voriconazole complexes containing perchlorate and nitrate counter anions indicated that these complexes showed good antifungal activity against the tested fungi and featured a higher activity against voriconazole-resistant Candida albicans strains compared to voriconazole (MIC90 = 0.0625–2.00 and 0.0625–64.00 mg/mL for silver(I) complexes and voriconazole, respectively; MIC90 is the lowest concentration which inhibits 90% of growth compared to the control) [28]. The antimicrobial tests of {[Ag(vcz)2]SbF6}n showed that this complex exhibited a significantly improved antifungal activity compared to the parent voriconazole with an increase of activity by five orders of magnitude against C. glabrata [29]. Furthermore, this complex inhibited the formation of C. albicans hyphae and biofilms at its subinhibitory concentration [29].
Inspired by the fact that variety in geometry and biological potential of silver(I) complexes with voriconazole antifungal drug strongly depend on the nature of silver(I) salt used for their synthesis, in the present work the synthesis and antimicrobial evaluation of new silver(I)-voriconazole complexes, {[Ag(vcz)(H2O)]CH3SO3}n (1), {[Ag(vcz)2]BF4}n (2) and {[Ag(vcz)2]PF6}n (3), are presented. These complexes have been experimentally and theoretically characterized and biologically evaluated as potential anti-Candida agents in comparison to the parent drug. In addition, a molecular docking study was conducted to investigate the inhibitory effect of the silver(I)-voriconazole complexes against the sterol 14α-demethylase enzyme cytochrome P450 enzyme (CYP51B) and compare it with the inhibitory activity of the parent drug.

2. Experimental

2.1. Materials and measurements

Silver(I) salts (AgCH3SO3, AgBF4 and AgPF6), voriconazole (vcz), ethanol, acetonitrile, dimethyl sulfoxide (DMSO) and deuterated dimethyl sulfoxide (DMSO‑d6) were purchased from the Sigma-Aldrich and Acros Organics. These chemicals were of analytical grade and used as received without further purification.
Elemental analyses of the synthesized complexes 1–3 for carbon, hydrogen and nitrogen were performed using a Perkin-Elmer 2400 Series II instrument. KBr pellet technique was used for recording the IR spectra on a Perkin Elmer Spectrum 2 spectrometer over the wavenumber range of 4000–450 cm−1 (abbreviations used for the intensity of the IR bands: br = broad, vs = very strong, s = strong, m = medium and w = weak). The 1H NMR spectra of silver(I) complexes and vcz, dissolved in 0.6 mL of DMSO‑d6, were recorded using a Varian Gemini 2000 spectrometer at 200 MHz at room temperature, while the time stability of the complexes 1–3 in DMSO‑d6 solution during 9 days was followed on a Bruker Ascend 400 MHz spectrometer. The 1H-15N HMBC spectra were measured for the silver(I) complexes in DMSO‑d6 at room temperature on a Bruker Avance III 600 MHz spectrometer. Chemical shifts, δ, are given in ppm (parts per million), while scalar couplings, J, are reported in Hz (Hertz). The splitting of protons is designated as follows: s, singlet; d, doublet; dd, doublet of doublets; dt, doublet of triplets; t, triplet; td, triplet of doublets; q, quartet, and m, multiplet. The UV–Vis spectra of silver(I) complexes in DMSO were recorded over the wavelength range of 1100–200 nm on a Shimadzu UV-1800 spectrophotometer at room temperature. The concentration of solutions of the complexes for measurement of their UV–Vis spectra was 8.3 × 10−4 М, 6.3 × 10−4 М and 7.7 × 10−4 М for 1–3, respectively. Moreover, the measurement of these spectra for the studied complexes was repeated after 24 and 48 h standing in the dark at room temperature to follow their solution behavior. The ESI-HRMS spectra in the positive mode were recorded after dissolving the silver(I) complex in CH3CN with an Agilent 62224 accurate mass spectrometer, using time of flight liquid chromatography/mass spectrometry.
Data from single crystals of silver(I) complexes (Fig. S1 and S2) were collected at 150 K on a SuperNova diffractometer with Atlas detector using CrysAlis software with monochromated Mo Kα (0.71073 Å) [30]. The initial structural models were solved with direct methods implemented in SHELXT using the Olex2 graphical user interface [31]. A full-matrix least-squares refinement on F2 magnitudes with anisotropic displacement parameters for all non‑hydrogen atoms using Olex2 or SHELXL-2018/3 was performed [31,32]. All non‑hydrogen atoms were refined anisotropically, while hydrogen atoms were placed at calculated positions and treated as riding on their parent atoms. Details on the crystal data, data acquisition and refinement are presented in Table S1. Mercury [33] was used for the preparation of the figures, presenting the crystal structures of the studied silver(I) complexes.

2.2. Synthesis of silver(I) complexes 1–3

Silver(I) complexes 1–3 were synthesized by modification of the method previously reported for the synthesis of silver(I) complexes with different azole ligands [29]. The solution of 0.25 mmol of vcz (87.3 mg) in 5.0 mL of ethanol was added slowly under stirring to the ethanolic solution (5.0 mL) containing an equimolar amount of the corresponding silver(I) salt (50.7 mg of AgCH3SO3 for 1, 48.7 mg of AgBF4 for 2 and 63.2 mg of AgPF6 for 3). A white precipitate was formed after the addition of the vcz ligand. The reaction mixture was further stirred in the absence of light at room temperature for 3 h, and then, the precipitate was filtered off and dissolved in acetonitrile. The obtained solution was left at room temperature and after 3–5 days, the white crystals of silver(I) complexes 1–3 suitable for single-crystal X-ray diffraction analysis, were formed. Yield: 87.0 mg (61%) for 1, 75.9 mg (68%) for 2 and 84.4 mg (71%) for 3.
Anal. calcd for 1 (C34H38Ag2F6N10O10S2; MW = 1140.60 g/mol): C, 35.80; H, 3.36; N, 12.28. Found: C, 35.68; H, 3.28; N, 12.32%. HRMS-ESI (CH3CN): m/z calcd for [C16H14AgF3N5O]+: 456.0201; found 456.0193. IR (KBr, ν, cm−1): 3418br (ν(O-H)), 3108m, 3037w (ν(Ctriazole–H) and ν(Car-H)), 2984w, 2941w (ν(C-H)), 1619s, 1597s, 1559w, 1526m, 1501s, 1464m, 1457m, 1436m, 1407s (ν(Car = Car) and ν(C=N)), 1293m (δ(O-H)), 1211vs, 1196vs (νas(SO3)), 1131m (ν(C-F)), 1039s (νs(SO3)), 867w, 846m, 824w, 773m (γ(Car-H)) and (γ(Ctriazole–H)), 622m (β(Car-F)). 1H NMR (200 MHz, DMSO‑d6): δ = 9.06 (d, J = 2.9 Hz, 1H, C18H), 8.87 (d, J = 2.2 Hz, 1H, C20H), 8.34 (s, 1H, C5H), 7.70 (s, 1H, C3H), 7.26 (dt, J = 6.9, 2.1 Hz, 1H, C9H), 7.19 (m, 1H, C12H), 6.93 (td, J = 8.3, 2.4 Hz, 1H, C10H), 6.02 (s, 1H, OH), 4.83 (d, J = 14.2 Hz, 1H, C6H), 4.37 (d, J = 14.3 Hz, 1H, C6H), 3.93 (d, J = 6.9 Hz, 1H, C14H), 1.10 ppm (t, J = 8.7 Hz, 3H, C15H). 15N NMR (61 MHz, DMSO‑d6): δ = 298.04 (N17/N19), 238.65 (N2/N4), 212.83 ppm (N1). UV–Vis (DMSO, λmax, nm): 291 (ε = 9.4 × 102 M−1 cm−1). S (DMSO, mol/L): 0.8767.
Anal. calcd for 2 (C32H28AgBF10N10O2; MW = 893.32 g/mol): C, 43.03; H, 3.16; N, 15.68. Found: C 42.87; H 3.04; N 15.82%. HRMS-ESI (CH3CN): m/z calcd for [C32H28AgF6N10O2]+: 807.1352; found 807.1324. IR (KBr, ν, cm−1): 3371br (ν(O-H)), 3139w, 3118w, 3074w, 3063w (ν(Ctriazole–H) and ν(Car-H)), 2994w, 2981w, 2939w, 2910w (ν(C-H)), 1618s, 1596s, 1525s, 1500vs, 1458m, 1440m, 1422m, 1407vs (ν(Car = Car) and ν(C=N)), 1293m, 1281m, 1272m (δ(O-H)), 1132m (ν(C-F)), 1059vs (ν(BF4)), 853s, 829w, 793w, 787w, 736w, 721w, 710w (γ(Car-H)) and (γ(Ctriazole–H)), 621m (β(Car-F)). 1H NMR (200 MHz, DMSO‑d6): δ = 9.05 (d, J = 2.9 Hz, 1H, C18H), 8.86 (d, J = 2.1 Hz, 1H, C20H), 8.29 (s, 1H, C5H), 7.66 (s, 1H, C3H), 7.27 (m, 1H, C9H), 7.18 (m, 1H, C12H), 6.92 (td, J = 8.5, 2.5 Hz, 1H, C10H), 6.01 (s, 1H, OH), 4.82 (d, J = 14.3 Hz, 1H, C6H), 4.35 (d, J = 14.2 Hz, 1H, C6H), 3.93 (d, J = 7.6 Hz, 1H, C14H), 1.15 ppm (t, J = 13.2 Hz, 3H, C15H). 15N NMR (61 MHz, DMSO‑d6): δ = 298.51 (N17/N19), 238.21 (N2/N4), 212.88 ppm (N1). UV–Vis (DMSO, λmax, nm): 291 (ε = 1.3 × 103 M−1 cm−1). S (DMSO, mol/L): 1.1194.
Anal. calcd for 3 (C32H28AgF12N10O2P; MW = 951.48 g/mol): C, 40.40; H, 2.97; N, 14.72. Found: C 40.28; H 2.91; N 14.81%. HRMS-ESI (CH3CN): m/z calcd for [C16H14AgF3N5O]+: 456.0201; found 456.0188. IR (KBr, ν, cm−1): 3415br (ν(O-H)), 3196w, 3147w (ν(Ctriazole–H) and ν(Car-H)), 2992w, 2980w, 2943w (ν(C-H)), 1618s, 1596s, 1524m, 1499s, 1422m, 1406s (ν(Car = Car) and ν(C=N)), 1292m, 1280m (δ(O-H)), 1132m (ν(C-F)), 838vs (ν(PF6)), 853s, 781w, 755w, 737w, 722w, 713w (γ(Car-H)) and (γ(Ctriazole–H)), 621m (β(Car-F)). 1H NMR (200 MHz, DMSO‑d6): δ = 9.06 (d, J = 2.9 Hz, 1H, C18H), 8.87 (d, J = 2.2 Hz, 1H, C20H), 8.31 (s, 1H, C5H), 7.68 (s, 1H, C3H), 7.22 (m, 2H, C9H, C12H), 6.94 (m, 1H, C10H), 6.01 (s, 1H, OH), 4.82 (d, J = 14.2 Hz, 1H, C6H), 4.36 (d, J = 14.3 Hz, C6H), 3.91 (t, J = 6.9 Hz, C14H), 1.12 (d, J = 7.1 Hz, 3H, C15H). 15N NMR (61 MHz, DMSO‑d6): δ = 298.14 (N17/N19), 245.43 (N2/N4), 212.24 ppm (N1). UV–Vis (DMSO, λmax, nm): 292 (ε = 1.1 × 103 M−1 cm−1). S (DMSO, mol/L): 0.3003.
For comparative purposes, NMR, IR and UV–Vis spectroscopic data for vcz are given in ESI.

2.3. Antifungal activity

According to the standard broth microdilution assay, minimum inhibitory concentration (MIC) values of the complexes 1–3, AgCH3SO3, AgBF4 and AgPF6, and vcz were determined [34]. The tested microorganisms were Candida strains: C. albicans ATCC 10231, C. parapsilosis ATCC 22019, C. krusei ATCC 6258 and C. glabrata ATCC 2001 (ATCC is American Type Culture Collection) with inoculums 1 × 105 colony forming units (cfu)/mL. The highest tested concentration was 250 μM (in DMSO). The MIC value was recorded after 24 h at 37 °C using a plate reader (Epoch Microplate Spectrophotometer, BioTek Instruments, Inc., USA).

2.4. Cell viability and proliferation assay

The antiproliferative activity was assessed using the MTT assay [35]. Cells were plated at 1 × 104 cells per well in a 96-well plate, cultured in RPMI (Roswell Park Memorial Institute) 1640 medium with 10% FBS (fetal bovine serum), and treated with serially diluted compounds (starting with 1000 μM) for 48 h. The MTT assay was conducted twice in four replicates, and results were presented as a percentage of DMSO-treated controls (set at 100%). IC50 values, representing the concentration inhibiting cell growth by 50%, were determined by measuring MTT reduction at 540 nm using a plate reader (Epoch Microplate Spectrophotometer, BioTek Instruments, Inc., USA). Cytotoxicity was expressed as the IC50 in comparison with the negative control (DMSO-treated cells).

2.5. Filamentation test on C. albicans ATCC 10231

The filamentation test on C. albicans ATCC 10231 was assessed using the solid medium [36]. Overnight culture of C. albicans grown at 37 °C was centrifuged for 10 min. The pellet was washed twice with sterile PBS (phosphate-buffered saline; Sigma Aldrich, Munich, Germany) and resuspended in 200 μL PBS. 2 μL of the PBS suspension was inoculated into the solid media with or without the tested compounds (subinhibitory concentration 0.5 × MIC) and incubated at 37 °C for 72 h. It is known that subinhibitory concentrations of antimicrobial compounds can produce significant effects in inhibiting the rate of microbial growth [37]. Moreover, some studies have highlighted the significant impact of subinhibitory concentrations of the antifungal compounds, including azoles, on clinical C. albicans isolates [38]. The morphological changes were analyzed using bright field microscopy (SMZ143-N2GG, Motic, Germany). DMSO was used as a control.

2.6. Anti-biofilm activity assessment on C. albicans ATCC 10231

The effect of the tested compounds on biofilm formation was determined on the C. albicans ATCC 10231 strain. Anti-biofilm assays were conducted using previously reported methodologies [22,39]. In the biofilm disruption assay, preformed biofilms (24 h at 37 °C) were incubated for 24 h with decreasing concentrations of the compound. Biofilm growth was quantified by crystal violet (CV) staining of adherent cells and estimated as absorbance at 530 nm on a plate reader (Epoch Microplate Spectrophotometer, BioTek Instruments, Inc., USA).

2.7. Ergosterol biosynthesis

After 18 h incubation at 37 °C on the rotary shaker (180 rpm), ergosterol levels in the untreated and treated cultures of C. albicans ATCC 10231 with 0.5 × MIC of vcz, silver(I) complexes and corresponding salts were assessed following a known procedure [40]. As was already mentioned, the subinhibitory concentrations of the antifungal azoles have a significant influence on clinical C. albicans isolates [38]. Ergosterol concentrations were determined by scanning the absorbance between 240 and 300 nm (Ultrospec 3300pro, Amersham Biosciences, USA).

2.8. Ergosterol biding assay – MIC value determination in the presence of ergosterol

To assess if the compound binds to the fungal membrane sterols, this experiment was performed according to the method previously described [41] with some modifications. The ergosterol was dissolved in DMSO (no >10% of final volume), and Tween 80 at 1%, in accordance with the desired concentration and volume. The formed emulsion was then homogenized, heated to augment the solubility, and diluted with the liquid culture medium. The MIC of vcz and silver(I) complexes against C. albicans was determined by using broth microdilution techniques previously described, in the presence and absence of exogenous ergosterol (Sigma-Aldrich) added to the assay medium, in different lines of the same microplate. A solution of tested compounds was serially doubly diluted with RPMI 1640 (volume 100 μL) containing additiona;l ergosterol at a concentration of 400 μg/mL. A volume of 10 μL of yeast suspension (0.5 McFarland) was added to each well. The same procedure was done for amphotericin B, whose interaction with membrane ergosterol is already known and which served as a control drug. The plates were incubated at 37 °C and read after 24 h of incubation. MIC was determined as the lowest concentration of the tested compound inhibiting the visible growth. This assay was carried out in duplicate.

2.9. Computational methodology

The calculations were conducted using the Gaussian16 software package in conjunction with the GaussView 6.0.16 program to provide a graphical representation [42]. The investigated structures were optimized using density functional theory (DFT) calculations, specifically employing the М06-2× functional and the 6–311 + G(d,p) basis set, which includes polarization and diffuse functions [43]. Additionally, the LANL2TZ(f) basis set, incorporating an effective core potential, was used for the silver atom [44]. Many available literature parameters showed the effectiveness of the M06-2× theoretical model in replicating the experimental geometric parameters [[45], [46], [47]]. The Conductor-like Polarizable Continuum Model (CPCM) was employed to approximate the effect of DMSO as a solvent on the geometrical parameters of the investigated compounds [48]. The change in the Gibbs free energy (ΔG0) associated with the formation or dissociation of the investigated complexes in DMSO was determined by subtracting the sum of the Gibbs energies of the products (G°products) from the sum of the Gibbs energies of the reactants (G°reactants), which were obtained following the optimization of the geometry of the investigated complexes (eq. 1) [49].
(1)

The equilibrium constant of the investigated reactions was determined utilizing the following equation (eg. 2; T = 298.15 K; R = 8.314 J K−1 mol−1) [49]:
(2)

The chemical shifts of the investigated compounds in 1H NMR have been determined using calculations conducted in DMSO as solvent. The Gauge Independent Atomic Orbital (GIAO) method was employed for these calculations [50]. The UV–Vis spectra of the investigated compounds were obtained utilizing Time-Dependent Density Functional Theory (TD–DFT) [51] within a DMSO as solvent. The Bond Critical Points (BCPs) of the silver(I)-voriconazole complexes were determined within the Quantum Theory of Atoms in Molecules (QTAIM) analysis using the Multiwfn software package [52].

2.10. Molecular docking simulation

The AutoDock 4.0 software program was utilized to perform a molecular docking simulation of the examined complexes to the sterol 14α-demethylase enzyme cytochrome P450 (CYP51B) as the selected target [53]. The three-dimensional crystal structure of CYP 450 format was obtained from the RCSB Protein Data Bank in PDB format (PDB code 4UYM, accession date: 20.11.2023; cocrystallized inhibitor: voriconazole) [54]. The software program utilized for protein preparation in the docking simulation was Discovery Studio 4.0 (BIOVIA Discovery Studio 2016) [55]. Water and other co-crystallized molecules are removed throughout the protein preparation process. The protoporphyrin IX containing Fe(II) (HEM 580) molecule is retained within the protein structure because various compounds interact with it and thus exert a significant inhibitory effect. The Hydrogen module inside the AutoDockTools (ADT) graphical interface was utilized to incorporate polar hydrogen atoms into proteins. Subsequently, the Kollman united atom partial charges were allocated to the proteins. The Lamarckian Genetic Algorithm (LGA) was employed to conduct protein-ligand rigid-flexible docking. The specific parameters used in this molecular docking simulation were more comprehensively described in prior studies [56,57]. The receptor was employed as rigid input receptor molecules, whereas the investigated compounds were utilized as flexible ligands. The ligand bonds were configured to be rotatable, and a uniform value of 14 active torsions was assigned to all investigated compounds. The search space of CYP51B was restricted to a grid box size of 70 × 70 × 70 Å following the XYZ dimensions: 135.543 × 196.724 × 3.934 by the utilization of the AutoGrid module.

2.11. Statistical analysis

The results are presented as mean ± standard deviation (SD). Statistical analysis was done by comparing means using a t-test (Two-Sample Assuming Equal Variances) and one-way ANalysis of VAriance (ANOVA, Single Factor), with Fisher’s Least Significant Difference (LSD) posthoc test. A p-value (probability value) ≤ 0.05 was considered statistically significant. Statistical analysis tests were performed in Microsoft Excel Spreadsheet Software by Data Analysis Tools add-in.

3. Results and discussion

3.1. Synthesis and characterization of the silver(I) complexes 1–3

The antifungal drug, voriconazole (vcz, Fig. 1), was used as a ligand for the synthesis of three new silver(I) coordination polymers, {[Ag(vcz)(H2O)]CH3SO3}n (1), {[Ag(vcz)2]BF4}n (2) and {[Ag(vcz)2]PF6}n (3) (Fig. 2). For the synthesis of these complexes, the method described in the Experimental section was used (vide supra). The single crystals of complexes 1–3 were formed after the precipitates from the reactions were recrystallized in acetonitrile and obtained solutions were kept at room temperature for 3–5 days. Elemental analysis, IR, UV–Vis and NMR (1H and 15N) spectroscopy, mass spectrometry, single-crystal X-ray diffraction analysis and DFT calculations were used to elucidate the structures of the silver(I) coordination compounds 1–3 both in solution and solid state.

Fig. 2. Crystal structure of complexes 1 (top left), 2 (top right) and 3 (bottom) with selected heteroatom labelling. Thermal ellipsoids are shown at a 35% probability level. Hydrogen atoms in all structures and the counterion in 1 are omitted for better clarity of presentation

3.1.1. Solid state studies

The structural database CSD (Cambridge Structural Database; accessed on 18.10.2023) contains 11 voriconazole metal complexes (4 for Cu, 4 for Ag, and 1 for Cd, Co, and Au). Both the tetrafluoroborate complex 2 and the hexafluorophosphate complex 3 are isostructural to the hexafluoroantimonate silver(I)-vcz complex recently published by our group [29], as well as a previously published trifluoromethanesulfonate (triflate) complex [18]. The multitude of sp2 nitrogen atoms in the vcz framework offers multiple binding sites for silver(I) ions. In the polymeric complexes 2 and 3, silver(I) ions are coordinated by four nitrogen atoms and adopt a distorted tetrahedral geometry (Fig. 2). In these complexes, a polymeric chain is formed by the vcz triazole ring acting as a bridging moiety between the two adjacent silver(I) ions. The remaining two sites in the silver(I) coordination sphere are occupied by one of the fluoropyrimidine nitrogen atoms of the neighboring molecule crosslinking the chains and triazole ring of the second (terminal) vcz ligand resulting in a net metal : vcz ratio of 1 : 2.
Moreover, herein we report a novel structure of a methanesulfonate salt of a silver(I)-vcz complex 1. Recrystallization from an acetonitrile solution yielded a complex in which the determined metal : vcz ratio is 1 : 1 (Fig. 2). Interestingly, the chain-like structure and the cross-linking pattern are formed in an identical manner to all other complexes, however the second (terminal) azole molecule is replaced by a water molecule. The structure also presents two small voids of 43 and 33 Å3 containing a residual electron density integrated into 12 and 11 electrons, respectively. Attempts to place water, ethanol or acetonitrile molecules at full or partial occupancy were unsuccessful and the solvent mask feature in Olex2 was used to remove the residual electron density from the structural model. A figure representing the voids is included in the ESI (Fig. S3).

The IR spectra of vcz and complexes 1–3 recorded in the wavenumber range 4000–450 cm−1 are shown in Fig. S4. The bands attributable to the coordinated vcz [58] and water [59], as well as those due to the corresponding CH3SO3−, BF4− and PF6− counter anions [[60], [61], [62], [63]], are observed in the IR spectra of the complexes. A broad band at ⁓3400 cm−1 originates from the O-H stretching vibrations of the vcz ligand of all three complexes, as well as from the coordinated water molecule of complex 1 [58,59]. The bands at ∼3100–3000 cm−1 and those at ∼2900 cm−1 are due to the Caromatic–H and Caliphatic–H stretching vibrations, respectively [59]. Moreover, the shift of the bands due to vibrations of the six- and five-membered aromatic rings ((ν(Car = Car) and ν(C=N))) for the complexes 1–3 compared to those of the uncoordinated vcz, observed in the range 1619–1406 cm−1 [59], confirms its coordination to the silver(I) ion. The presence of CH3SO3−, BF4− and PF6− counter anions in the crystal lattice of 1–3, respectively, can also be confirmed from the IR spectra of these complexes. Thus, very strong bands due to the asymmetric and symmetric stretching vibrations of the –SO3 group of the methanesulfonate anion are observed at 1196 and 1039 cm−1, respectively, in the IR spectrum of 1, whereas these values are in accordance with those for methanesulfonic acid itself [60]. Regarding 2 and 3, the presence of strong bands at 1059 and 838 cm−1 attributed to the stretching vibrations of the BF4− and PF6−, respectively, is a consequence of their counter anion role [[61], [62], [63]]. This agrees with the IR spectroscopic behavior of the previously reported {Ag(1,6-naph)(H2O)}n and Ag2(1,8-naph)2(H2O)1.22 coordination compounds (1,6- and 1,8-naph is 1,6- and 1,8-naphthyridine) [64,65].

3.1.2. Solution behavior

The UV–Vis spectra of vcz and 1–3 were recorded in DMSO at room temperature (Fig. S5). As was previously reported for UV–Vis spectroscopic features of silver(I) complexes with different antifungal azoles [29], the absorbance peaks at λmax ∼ 290 nm can be assigned to the intra-ligand charge transfer transitions. In comparison to the absorbance peak of the uncoordinated vcz (λmax = 256 nm), the corresponding peaks for the silver(I) complexes show bathochromic red shifts (Fig. S5). The measurements of the UV–Vis spectra of 1–3 were repeated after standing of the corresponding solutions at room temperature in the absence of light for 48 h to study the stability of the investigated compounds in the solution. As can be seen in Fig. S5, only a slight decrease in the absorption intensity between 3 and 7% was observed for 1–3, suggesting that the vcz ligand remains coordinated for the silver(I) ion during this time.

The positive ion ESI mass spectra for 1–3 in acetonitrile solution show major peaks centered at m/z = 456.0193, 807.1324 and 456.0188, respectively. These m/z values are consistent with those calculated for [Ag(vcz)]+ (m/z = 456.0201 for 1 and 3) and [Ag(vcz)2]+ (m/z = 807.1352 for 2) cations, indicating that mononuclear cationic silver(I) species are present in solution under ESI-MS conditions.

Comparative 1H NMR spectra of complexes 1–3 and the vcz ligand, recorded in DMSO‑d6, are shown in Figs. S6-S8. These spectra were assigned based on the 1H NMR data obtained for the previously synthesized complexes, {[Ag(vcz)2]ClO4}n [28], {[Ag(vcz)2]NO3}n [28] and {[Ag(vcz)2]SbF6}n [29]. As can be seen in Figs. S6-S8, the 1H NMR spectra of 1–3 show only minor differences in the chemical shifts of protons compared to those of the uncoordinated vcz ligand. The most significant Δ(1H)coord coordination shifts (calculated with respect to vcz) of +0.10, +0.05 and + 0.07 ppm for complexes 1–3, respectively, were found for the H5 proton bound to the triazole carbon atom, which is in the vicinity of the metal-coordinated nitrogen, indicating that vcz is still coordinated to the silver(I) ion even though the investigated complexes were dissolved in DMSO‑d6. Moreover, vcz remains coordinated to the silver(I) ion after 9 days standing at room temperature (Figs. S9-S11).

In the case of nitrogen-donor ligands, 15N NMR spectra can provide a valuable information regarding metal coordination [66]. However, the observed coordination effect in 1H NMR spectra of the complexes (i.e. only minor differences in the chemical shifts compared to vcz) was in accordance with the change in 15N NMR coordination shifts (Δδ15Ncoord = δ15Ncomplex – δ15Nligand) (Fig. S12). The triazole atoms N2 and N4 are the most affected nitrogen atoms by voriconazole coordination to the silver(I) ion with the coordination shifts Δδ15Ncoord = −13.69, −14.13 and − 6.91 ppm for 1–3, respectively (Fig. S12).

3.2. DFT calculations

To gain a comprehensive understanding of the structures and properties of the synthesized complexes 1–3 in DMSO, we performed an investigation into the thermodynamics of possible reactions of formation and dissociation, commencing from their X-ray-determined structures. To achieve this objective, the complex tetrahedral geometries of 1–3 in DMSO were optimized and shown in Fig. S13, while the optimized linear geometries of [Ag(vcz)2]+ complex units 1a – 3a, respectively, are shown in Fig. 3.

Fig. 3. Optimized geometries of linear [Ag(vcz-N2)(vcz-N19)]+ (1a), [Ag(vcz-N2)(vcz-N4)]+ (2a) and [Ag(vcz-N4)2]+ (3a) complexes in DMSO evaluated by M06-2× functional in combination with 6–311 + G(d,p) basis set for C, H, O, N and F atoms and LANL2TZ(f) basis set for Ag atom. Dashed lines represent hydrogen bonds. Legend: C (grey), H (white), O (red), N (dark blue), Ag (light blue). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
The ΔrG (298 K) values for the reactions of the formation (ΔGf0) and dissociation (ΔGd0) of the silver(I)-vcz complexes in DMSO were determined by calculating the difference between the Gibbs free energies of the products and reactants (eq. 1) using the М06-2× functional in combination with the 6–311 + G(d,p) basis set for all atoms and LANL2TZ(f) basis set for Ag atom (Table 1).
Table 1. Gibbs free energy changes (T = 298.15 K, kJ/mol) and constant of formation (ΔGf0, kJ /Kf) and dissociation (ΔGd0/Kd) of silver(I)-vcz complexes calculated by M06-2× functional in combination with 6–311 + G(d,p) basis set for C, H, O, N and F atoms and LANL2TZ(f) basis set for Ag atom.

Initially, an investigation was conducted on the formation of complexes 1a – 3a in the reaction between silver(I) salts, AgCH3SO3, AgBF4 and AgPF6, respectively, and vcz in a stoichiometric ratio of 1 : 2. The results presented in Table 1 demonstrate the significant exergonic ΔGf0 values, suggesting a favouring the formation of linear [Ag(vcz)2]+ complexes, in which the nitrogen atoms of aromatic triazole (N2 for 1a, N2 and N4 for 2a, and two N4 atoms for 3a) and pyrimidine (N19 for 1a) rings are involved in coordination to the silver(I) ion. Moreover, the complexation reactions are characterized by positive values of the formation constants (> 102).

On the other hand, the formation of complexes 1a – 3a was initiated by dissociation of the tetracoordinate [Ag(vcz-N2)(vcz-N4)(vcz-N19)(H2O)]+ (for 1a) and [Ag(vcz-N2)(vcz-N4)2(vcz-N19)]+ (for 2a and 3a) complexes (Table 1). The thermodynamic favourability of dissociation of two vcz ligands in [Ag(vcz-N2)(vcz-N4)2(vcz-N19)]+ or vcz and water molecules in [Ag(vcz-N2)(vcz-N4)(vcz-N19)(H2O)]+ resulting in the formation of linear species 1a – 3a was determined based on the values of the ΔGd0 < 0 kJ/mol and Kd > 1. Moreover, the stability of these linear complexes in the presence of DMSO is indicated by positive ΔGd0 values and a dissociation constant Kd lower than 1, suggesting that the studied complexes 1–3 in DMSO are represented as linear 1a – 3a species (Fig. 3 and Table 1).

To determine the geometrical parameters of the silver(I)-vcz complexes in DMSO, chemical shifts in 1H NMR spectra were simulated in relation to TMS (tetramethylsilane), which served as an internal standard (Table S2). The high correlation factor (R) between experimental and simulated chemical 1H NMR shifts (> 0.991), along with the low values of mean absolute error (MAE) (< 0.60), strongly confirms the presence of linear 1a – 3a complexes in DMSO. Also, this is an additional confirmation that the used theoretical model accurately characterizes the molecular structures of the synthesized compounds. The utilization of the implicit solvent model and the absence of direct interactions with solvent molecules result in slightly elevated simulated chemical shifts of aromatic protons, particularly H9, H18, and H20.

Additionally, the presence of complexes 1a – 3a in DMSO was confirmed by comparing the experimental and simulated UV–Vis spectra. Fig. S14 presents comparison of experimental (black line) with theoretical (red line) UV–Vis spectra of 1–3 and 1a – 3a complexes, respectively, displayed as a function of absorbance or oscillator strength against wavelength (nm). Table S3 shows the quantitative experimental and theoretical values of the most significant electronic transitions between HOMO (Highest Occupied Molecular Orbital) and LUMO (Lowest Unoccupied Molecular Orbital) orbitals. The experimental UV–Vis spectra of the silver(I)-vcz complexes exhibit a single prominent maximum at wavelengths of 291 (1 and 2) and 292 (3) nm (Table S3). Within the simulated spectra, there is an observed peak resulting from two dominant electronic transitions. The occurrence of peaks at wavelengths 292 (1a, f = 0.230), 298 (2a, f = 0.115), and 295 (3a, f = 0.230) nm in theoretical spectra can be attributed to electronic transitions between the HOMO → LUMO (∼60%) (Figs. S15 – S17). Conversely, the observed peaks at somewhat higher wavelengths: 297 (1a, f = 0.439), 307 (2a, f = 0.231), and 295 (3a, f = 0.230), accompanied by higher oscillator strength values, can be attributed to HOMO-8 and LUMO+1 transitions (Figs. S15 – S17). The difference between experimental and theoretical values can be attributed to the calculation of electronic transitions in the polarizable continuum, without explicit modeling of specific interactions with solvent molecules. In short, the investigated complexes possess a significant number of reactive positions that could establish strong intermolecular interactions with solvent molecules and change the position of the electronic level.

The present study employed QTAIM analysis to investigate the notable intramolecular interactions that play a role in stabilizing the structures of complexes 1a – 3a in DMSO. Fig. S18 illustrates the significant bond critical points (BCP, 1–4), whereas Table S4 presents the quantitative values of significant QTAIM descriptors: electron density (ρ(r)), electron density Laplacian (∇2ρ(r)), total electron energy density (H(r)), potential ratio (V(r)) and kinetics (G(r)) of electron energy density and interaction energy (Eint). The critical points BCP-1 and BCP-2 of complexes 1a – 3a have negative values for H(r) and ratio –(G(r)/(V(r)), falling within the range of 0.5 < −(G(r)/(V(r)) < 1. Moreover, these bonds exhibit elevated values of ρ(r) and ∇2ρ(r), suggesting that points BCP-1 and BCP-2 can be classified as coordination bonds with a covalent character. The energy (Eint, kJ/mol) associated with these interactions varies between −105.4 and − 222.7 kJ/mol. In contrast, critical points BCP-3 and BCP-4 exhibit lower values of ρ(r) and ∇2ρ(r), accompanied by slightly negative values of H(r) with a range of –(G(r)/(V(r)) where 0.5 < −(G(r)/(V(r)) < 1. These findings indicate that the observed interaction can be classified as a strong hydrogen bond. The Eint values corresponding to these interactions exhibit a range spanning from −25.6 to −31.6 kJ/mol.

3.3. Evaluation of the antifungal activity of silver(I)-vcz complexes

In the present study, we have assessed the effects of the complexation of clinically used voriconazole with silver(I) ion on its in vitro antifungal properties. In Table 2, the MIC values (μM) of complexes 1–3, vcz and silver(I) salts against Candida strains were presented and compared with their IC50 values (μM) against healthy human fibroblasts MRC-5 to determine the selectivity index (SI) of the complexes.

Table 2. Minimum inhibitory concentrations (MIC, μM) of complexes 1–3, vcz ligand and silver(I) salts against Candida strains in comparison to their IC50a values against healthy human fibroblasts MRC-5 (μM).

a

IC₅₀ is the concentration of a compound that inhibits 50% of cell growth after treatment of 48 h; ^bStandard deviation values were between 0 and 2%.

Very pronounced antifungal activity was observed for all tested compounds (Table 2). Compared to the clinically used vcz, the corresponding silver(I) complexes 1–3 showed significantly increased antifungal activity. Thus, complex 1 showed 256- and 11440-fold increased activity against C. albicans ATCC 10231 and C. glabrata ATCC 2001, respectively, compared to the parent organic compound. The effect of complex 2 was manifested trough 199-fold (C. albicans ATCC 10231), 15-fold (C. parapsilosis ATCC 22019), 11-fold (C. krusei ATCC 625) and 9533-fold (C. glabrata ATCC 2001) improvement compared to vcz. The remarkable improvement of the anti-Candida potential of vcz was also observed after the incubation of Candida spp. with complex 3 (105-, 15-, 11- and 545-fold against C. albicans ATCC 10231, C. parapsilosis ATCC 22019, C. krusei ATCC 6258 and C. glabrata ATCC 2001, respectively).

The obtained results are in agreement with those previously reported for the vcz-containing silver(I) complexes {[Ag(vcz)2]ClO4}n, {[Ag(vcz)2]NO3}n [28] and {[Ag(vcz)2]SbF6}n [29], suggesting that the strategy based on the combination of the clinically used antifungal drug voriconazole with the silver(I) ions could be successfully used for the development of a new anti-Candida therapy. Moreover, similar activity against the Candida species (C. albicans, C. krusei and C. glabrata) was previously manifested by [ZnCl2(vcz)2] complex, being also more effective than the parent vcz drug [67]. The same is true for copper(II) complexes with this antifungal agents, [Cu(CH3COO)2(vcz)2(H2O)]·2H2O, [Cu(NO3)2(vcz)2], and Cu2(vcz)2(H2O)82·4H2O, whereby the [Cu(CH3COO)2(vcz)2(H2O)]·2H2O complex was the most efficient one [68]. On the other hand, [AuCl3(vcz)] complex [69] was less effective against the same Candida species than the presently studied complexes 1–3, emphasizing the role of the metal ion.

The silver(I) salts used for the synthesis of 1–3 showed remarkable anti-Candida potential, and were more active than vcz, especially against C. albicans ATCC 10231 and C. glabrata ATCC 2001 strains (Table 2). Nevertheless, in most cases, their anti-Candida activity is lower than that of the corresponding silver(I) complex. However, the application of simple silver(I) salts for the treatment of microbial infections is limited due to their fast dissociation, which leads to the precipitation of AgCl under physiological conditions [70].

In terms of antiproliferative effect, compared to the vcz, silver(I) complexes 1–3 expressed a more pronounced effect on healthy human fibroblasts MRC-5, with the determined IC50 values being significantly lower than that of vcz (50-, 50- and 37-fold for 1–3, respectively; Table 2). Nevertheless, in the case of C. albicans and C. glabrata growth inhibition, the SI values (Table S5) determined for 1–3 were considerably higher in comparison to vcz.

Considering the notable anti-Candida activity with MIC ranging from 0.02 to 1.05 μM and a good therapeutic potential of the new silver(I)-vcz complexes, their effect on the invasion potential of C. albicans was further studied.

3.3.1. Filamentation test and anti-biofilm activity assessment on Candida albicans ATCC 10231

Our previous research has shown that the presence of metal ions, including Ag(I), Mn(II), Cd(II), Cu(II), and Zn(II) can lead to inhibition of the cellular differentiation of Candida [[71], [72], [73]]. Therefore, the effect of silver(I) complexes 1–3 and vcz, applied at concentrations corresponding to the previously determined MIC values, on C. albicans ATCC 10231 hyphae formation was analyzed using light microscopy (Fig. 4). As can be seen from Fig. 4, all tested compounds, at their MIC values, completely prevent C. albicans ATCC 10231 filamentation. These results agree with the previous studies showing that even vcz alone can inhibit the transformation of C. albicans from spores to invasive hyphae [74,75].

Fig. 4. Filamentation of C. albicans ATCC 10231 strain treated with MIC of 1–3 and vcz after 24, 48 and 72 h compared to the DMSO control.

C. albicans is associated with biofilm formation in immunocompromised and medically compromised patients, and it is well-established that biofilm formation is a major virulence factor during candidiasis [76]. In this study, it was found that complexes 1–3 inhibited the formation of biofilms of C. albicans ATCC 10231 at 0.5 × MIC in a high percentage of 79, 80, and 77%, respectively (Table S6). Compared to vcz (62%), the biofilm formation inhibition in the presence of the silver(I) complexes is higher, emphasizing the role of silver(I) ions. On the other hand, no influence of the complexes and vcz was observed against the already formed biofilms of C. albicans (data not shown).

3.3.2. Ergosterol biosynthesis

Taking into consideration the fact that azoles, in general, are blocking ergosterol synthesis by inhibiting the cytochrome P450, more precisely by inhibiting lanosterol 14α-demethylase, which converts lanosterol to ergosterol in fungal cellular membranes [77,78], we investigated if complexes 1–3 possess the same or similar activity to that of vcz. However, under our experimental conditions, the inhibition of the ergosterol biosynthesis in the presence of vcz was not significant in comparison to all three silver(I) complexes (Fig. 5). Moreover, all silver(I) salts used for the synthesis of complexes showed similar levels of ergosterol biosynthesis inhibition like the complexes, showing that the silver(I) ion is probably the main reason for this effect. These results are in line with those from the previous study [79], which showed that the exposure of C. albicans to metal nanoparticles, especially AgNPs used as carrier systems for conventional antifungals such as vcz, initiates the reduction of the ergosterol levels. Certainly, the lack of effect of vcz on ergosterol led us to investigate the ability of vcz and silver(I) complexes 1–3 to interact with ergosterol observing the effect of exogenous ergosterol on the tested compounds (Table S7). As can be seen, the silver(I) complexes 1–3 and vcz displayed changes in MIC and 0.5 × MIC values (results not shown) in medium with and without exogenously supplemented ergosterol, indicating that the mechanism of their action involves interaction with ergosterol. It could be noticed that the addition of exogenous ergosterol prevented the binding of the investigated compounds to ergosterol in the fungal membrane, since an increase of MIC values (for at least 1.2-fold) was observed for all of them. Moreover, only an increased product concentration in the growth medium might assure interaction with ergosterol in fungal membranes in accordance with the previously published results [41,80].

Fig. 5. Effect of subinhibitory concentration (0.5 × MIC) of complexes 1–3 and vcz on ergosterol level in C. albicans ATCC 10231 strain.

3.4. Molecular docking

Several groups of compounds, including azoles, pyridines, and pyrimidines, have been found to exhibit inhibitory effects on the sterol 14α-demethylase enzyme cytochrome P450 (CYP51B, EC 1.14.13.70). This enzyme catalyses a series of oxidative processes that remove 14α-methyl groups from cyclized sterol precursors and is involved in the biosynthesis of ergosterol, a vital constituent of fungal membranes [54,81]. The depletion of ergosterol has a significant impact on the structure of the membrane and several of its functions. Interaction of antifungal azoles with CYP51 has been studied in detail and it was postulated that triazoles form van der Waals contact within hydrophobic pocket of the active site [82]. Several studies have reported that the nitrogen atoms of the azole ring coordinate to the iron(II) ion within the heme cofactor [[83], [84], [85]]. If in metal complexes, the azole nitrogen atoms are involved in coordination to the silver(I) ions, binding to iron(II) ions in heme is prevented. It seems reasonable that such complexes thus inhibit the fungal growth differently than free azole drugs. Interestingly, we have also previously observed that the binding of ruthenium(II) to azole antifungals, did not result in a loss of activity [86].

The inhibitory capacity of vcz is demonstrated through its coordination with the HEM 580 iron of P450 [54,87,88]. Additionally, the unbound part of the inhibitor molecule establishes interactions with the protein component that predominantly determines the extent of inhibition [54,87,88].

The investigation of the inhibitory action of the silver(I)-vcz complexes against CYP51B, and its comparison with the inhibitory activity of vcz, holds significance in comprehensively elucidating the potential mechanisms of their inhibitory activity. Considering this rationale, a comprehensive investigation utilizing molecular docking simulation was undertaken, and the results have been delineated in Table 3. In this study, complexes 1a – 3a were used, since it was confirmed that silver(I)-vcz complexes exist in this form in a polar medium.

Table 3. Thermodynamic parameters (ΔGbind free energy binding, Ki constant of inhibition, ΔGtotal final total internal energy, ΔGtor torsional free energy, ΔGunb unbound system’s energy, ΔGelec electrostatic energy and is the sum of dispersion and repulsion (ΔGvdw), hydrogen bond (ΔGhbond), and desolvation (ΔGdesolv) energy) predicted for most favorable conformation of complexes 1a – 3a and vcz in the active site of cytochrome P450 sterol 14α-demethylase (CYP51B).

The thermodynamic parameters in Table 3 demonstrate that all the compounds examined exhibit superior inhibitory efficacy compared to vcz, being in line with our experimental results (vide supra). Based on the value of the free energy binding (ΔGbind) and constant of inhibition (Ki), the reactivity of the investigated compounds decreases in a series: CYP51B-3a > CYP51B-1a > CYP51B-2a > CYP51B-vcz.

Figure 6 displays the most stable conformations of the investigated compounds in the active site of the CYP51B enzyme. These compounds are positioned in a manner that the triazole ring and the partial negative fluorine atoms of the aromatic ring form contacts with the heterocyclic protoporphyrin IX ring (HEM 580) and the central metal ion, which is co-crystallized within the protein structure.

Fig. 6. The most favorable docking position of 1a – 3a and vcz in the active site of cytochrome P450 14α-sterol demethylase (CYP51) (PDB code: 4UYM). Bonds in the complexes are represented as grey sticks. Different colours on the sticks indicate different atoms: C (grey), N (dark blue), O (red), F (light blue) and Ag (blue). For clarity, the rest of the protein structure is omitted. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

An extensive examination of the interactions between the investigated compounds and amino acid residues is crucial for a comprehension of their potential mechanisms of action and, in general, for rational drug design. Fig. 7 illustrates the specific interactions that lead to the stability of the investigated compounds within the active site of the enzyme. Conventional hydrogen bonds are a significant form of interaction that contributes to the overall stability of the complex and its potential inhibitory activity. The vcz molecule forms two hydrogen bonds with SER 311 amino acid residue (2.68 and 3.73 Å), whereas complex 1a forms bifurcated hydrogen bonds involving the partly negative fluorine atom of the aromatic ring (2.01 Å) and the –OH group (1.80 Å) with the TYR 136 amino acid residue. Complex 3a forms hydrogen bonds with the –OH group of amino acid residues TYR 68 (1.81 Å) and TYR 122 (2.84 Å) by interacting with the partly positive nitrogen atoms of the triazole ring. The amino acid residues VAL 121 and LEU 91 form specific halogen interactions between the σ-hole (positive electrostatic potential) of fluorine atoms of 5-fluoropyrimidine and the aromatic ring of complexes 1a – 3a via partly positive nitrogen atoms. Halogen interactions have been extensively studied and have important applications in the fields such as crystal engineering and supramolecular chemistry [89,90]. Halogen atoms are frequently utilized in rational drug design to enhance the binding affinity, membrane permeability, and metabolic stability of medicines [91,92]. The stability of the silver(I) complexes in the active site of the CYP51B enzyme is significantly affected by hydrophobic contacts, namely π-sigma (TYR 122, ILE 373, ILE 377 and LEU 503), π-alkyl (ALA 307, ILE 373, ILE 376, ILE 377 and LEU 503), and π-π stacked (TYR 68, VAL 121, TYR 136, HIS 374 and PHE 504) interactions.

Fig. 7. 2D representation of interactions between 1a – 3a, vcz and amino acid residues in the active site of cytochrome P450 14α-sterol demethylase (CYP51) (PDB code: 4UYM) with interatomic distance obtained after molecular docking. HEM 580 is a co-crystallized protoporphyrin IX containing Fe(II). Different colours indicate different types of interactions (legend).

While a particular part of the molecule engages with amino acid residues of CYP51B, the 5-fluoropyrimidine ring of the examined compounds interacts with the heterocyclic protoporphyrin IX ring (HEM 580) co-crystallized in the protein structure (Fig. 8). The alignment of the aromatic 5-fluoropyrimidine ring of complex 3a with protoporphyrin results in the formation of four π-π stacked interactions with the imidazole rings of HEM 580. Furthermore, the Fe(II) ion with a partial positive charge forms a characteristic π-cation interaction with the aromatic 5-fluoropyrimidine ring (3.21 Å). Complex 3a in the active site of CYP51B is further stabilized by the porphyrin ring through π-lone (2.94 Å) and halogen bonds (2.82 Å). Complexes 1a and 2a form a limited number of interactions with HEM 580 (Fig. 8). The provided data exhibit a correlation with ΔGbind and Ki values, indicating that 3a demonstrates the highest inhibitory activity against CYP51B.

Fig. 8. Isolated 3D representation of the interactions between the 5-fluoropyrimidine ring of complexes 1a – 3a and protoporphyrin IX containing Fe(II) (HEM 580). Different colours indicate different types of interactions (legend).

4. Conclusions

In this study, three novel silver(I) complexes with the antifungal agent voriconazole (vcz) as ligand were synthesized, structurally characterized, and tested for their in vitro anti-Candida potential. The reactions between equimolar amounts of AgX (X = CH3SO3−, BF4− and PF6−) and vcz in ethanolic solution led to the formation of the silver(I) coordination polymers, {[Ag(vcz)(H2O)]CH3SO3}n (1), {[Ag(vcz)2]BF4}n (2) and {[Ag(vcz)2]PF6}n (3). Interestingly, although all reactions were performed under the same experimental conditions, the metal : ligand molar ratio in complex 1 is 1 : 1, while vcz coordinates to silver(I) ions in a 1 : 2 metal : ligand ratio in 2 and 3, indicating that the counter anion of the starting silver(I) salt has an important effect on the solid state structure of the silver(I) complex formed. DFT calculations have shown that complexes 1–3 exist as linear [Ag(vcz-N2)(vcz-N19)]+ (1a), [Ag(vcz-N2)(vcz-N4)]+ (2a) and [Ag(vcz-N4)2]+ (3a) species in DMSO, all of which differ in the donor atoms of vcz coordinated to the silver(I) ion. Comparison of the anti-Candida activity of the complexes 1–3 with that of vcz clearly shows that silver(I) complexes exhibit significantly higher antifungal activity against all strains tested, with favorable selectivity indices in the range from 19.3 to 1150. The anti-Candida potential of the silver(I)-voriconazole complexes also depends on the counter anion. Thus, complex 1 with CH3SO3− is the most active against C. parapsilosis and C. krusei, while its activity is lower against the two remaining Candida species (C. albicans and C. glabrata) in respect to the complexes 2 and 3 with BF4− and PF6− counter anioins, respectively. Furthermore, the synthesized silver(I) complexes have a significant effect on the virulence factors of C. albicans and prevent the formation of hyphae and biofilms of this fungus. The experimental results, as well as the molecular docking analysis results, show that the silver(I)-vcz complexes have a remarkable inhibitory effect on the CYP51B enzyme, which is essential for the production of ergosterol, a key component of the fungal membrane, being significantly higher than that of the parent organic drug. In view of all these results, the reported silver(I) complexes with voriconazole should be further investigated for possible use in therapy against lethal Candida infections.

References