Integrative computational-experimental discovery of α-hederin as a multi-mechanistic, low-toxicity antifungal agent targeting Candida albicans CYP51

Graphical abstract

Highlights

• •Computational screening yields α-hederin and elemenin as antifungal candidates.
• •α-Hederin shows high antifungal activity and low cytotoxicity.
• •It inhibits C. albicans hyphal/biofilm formation and cell surface hydrophobicity.
• •In murine model, α-hederin surpasses fluconazole in reducing fungal burden.
• •Molecular dynamics show α-hederin forms stable CYP51 complex with multiple bonds.

Abstract

Introduction

Infections caused by pathogenic fungi such as Candida albicans have led to a continuous increase in disease morbidity and mortality, underscoring the urgent need for safer and more effective antifungal therapies. Current antifungal drugs are limited by toxicity and resistance, highlighting the necessity for innovative discovery strategies.

Objectives

The aim of this study was to identify novel antifungal compounds using a integrative computational approach targeting C. albicans CYP51 (14-α-sterol demethylase). Specifically, we sought to screen the Traditional Chinese Medicine Systems Pharmacology (TCMSP) database for potential antifungal candidates, evaluate their therapeutic potential, and establish a computational-experimental framework to expedite antifungal drug development.

Methods

We employed an integrative computational approach-targeting Candida albicans CYP51 (14-α-sterol demethylase) via molecular docking and pharmacophore modeling-to screen the TCMSP database. The two natural small molecules obtained were subjected to in vivo and in vitro antifungal experiments, and their antifungal mechanisms were analyzed through molecular docking and molecular dynamics simulation.

Result

This yielded two novel antifungal candidates: α-hederin and elemenin, with minimum inhibitory concentrations (MICs) of 32 μg/mL and 16 μL/mL against C. albicans, respectively. α-Hederin demonstrated superior therapeutic potential, showing low cytotoxicity in 293 T, Raw264.7, and KB cells. It effectively inhibited hyphal formation, biofilm formation, and cell surface hydrophobicity in vitro. In a murine oral candidiasis model, α-hederin outperformed fluconazole by reducing fungal burden, inhibiting hyphal invasion, and preventing tongue adhesion. Molecular dynamics simulations revealed that α-hederin forms a more stable complex with CYP51 than fluconazole, engaging additional hydrogen bonds, alkyl interactions, and carbon-hydrogen bonds.

Conclusion

This study advances antifungal drug discovery by validating CYP51 as a high-value target for structure-guided screening, identifying α-hederin as a low-toxicity, multi-mechanistic antifungal agent and establishing a computational-experimental framework for rapid antifungal development.

Introduction

Invasive fungal infections caused by Candida and Aspergillus species pose severe threats to human health, particularly among immunocompromised patients. Globally, an estimated 150 million cases of severe fungal infections and 1.5 million related deaths are reported annually [1], with candidiasis being the predominant contributor. Among Candida species, Candida albicans is the most prevalent pathogen, responsible for 50 % of all candidiasis cases. C. albicans colonizes human mucosal surfaces such as the oral cavity, vagina, and gastrointestinal tracts [2], leading to oral candidiasis and vulvovaginal candidiasis. Oral infections disrupt the oral microenvironment and may disseminate systemically via digestive or respiratory routes, exacerbating disease severity [3].

The pathogenicity of C. albicans primarily stems from its ability to produce virulence factors [4].The fungus adheres to host epithelial cells through surface adhesins [5], proliferates at colonization sites, and secrets aspartic proteases and phospholipases to degrade extracellular matrix proteins, destroying host cell membrane integrity and promotes invasion [6]. Additionally, C. albicans undergoes a morphological transition from yeast to hyphal forms during infection. Hyphae exhibit upregulated expression of virulence-associated genes, enhancing tissue penetration and pathogenicity. Biofilm formation and cellular hydrophobicity further contribute to its resilience and invasiveness [7]. Therefore, inhibition of C. albicans growth and virulence factor production is critical for effective infection control.

Current antifungal therapies-primarily azoles (e.g., fluconazole) and polyenes (e.g., amphotericin B)-act by disrupting ergosterol biosynthesis or membrane integrity. Azoles inhibit CYP51 (14-α-sterol demethylase), a key enzyme in ergosterol synthesis, thereby increasing membrane permeability [8]. However, their clinical utility is limited by toxicity [9] and rising drug resistance [10,11]. These challenges underscore the urgent need for novel antifungals with improved safety and efficacy.

Natural small molecules, derived from plants and herbs, offer structurally diverse and low-toxicity alternatives [12]. Compounds such as carvacrol and geraniol inhibit C. albicans by disrupting cell membranes or suppressing virulence factors like biofilm formation [13,14]. Yet, traditional discovery methods are inefficient due to random screening and unclear mechanisms. Computer-aided virtual screening of natural small-molecule compounds enables simultaneous targeting of multiple biological pathways, offering high efficiency, well-defined mechanisms of action, and rapid identification of bioactive candidates requiring only experimental validation. Modern computational approaches-including homology modeling, molecular docking, pharmacophore modeling, and molecular dynamics simulations-significantly reduce the randomness of traditional drug discovery and enhance the translational potential of screened compounds. For instance, the TCMSP database, which catalogs 499 medicinal herbs, 29,384 natural compounds, and 13,133 drug-like molecules alongside critical ADMET (absorption, distribution, metabolism, excretion, and toxicity) parameters, serves as a powerful resource to accelerate antifungal drug discovery.

Current virtual screening strategies for antifungals fall into two categories: target-based molecular docking, which focuses on protein–ligand interactions, and ligand-based pharmacophore modeling, which identifies structural motifs (e.g., hydrogen-bond donors/acceptors, hydrophobic regions, and charged domains) to generate 3D pharmacophore models. These models are validated using test sets to assess predictive accuracy, enabling the selection of compounds with optimal antifungal activity [15]. For example, Dong et al. developed pharmacophore models for two antifungal targets, squalene epoxidase (SE) and CYP51, via the Hip-Hop algorithm. By analyzing ligand binding conformations within the CYP51 active site, they designed novel inhibitors through fragment matching, yielding compounds with potent antifungal effects [16].

Notably, both target- and ligand-based approaches rely on fungal-specific proteins to minimize off-target toxicity in humans. Ergosterol, a fungal membrane sterol absent in human cells, represents a prime target. Its biosynthetic enzyme CYP51 (particularly the erg11-encoded isoform) is inhibited by azoles through coordination with the heme cofactor in its active pocket, blocking oxygen binding and disrupting ergosterol synthesis [17]. Crucially, the β5 region of fungal CYP51 exhibits low homology to human orthologs, making it an ideal antifungal target. Prasher et al. synthesized indole derivatives (e.g., Compound 1) that inhibit C. albicans by forming hydrogen bonds with CYP51 residues (T763, S644, H759, D676), thereby suppressing ergosterol production [18]. Similarly, Singh et al. designed indole-based Compound 2, which interacts with CYP51 residues Y145, T135, I379, and M378, demonstrating both antifungal efficacy and mechanistic insights [19]. These findings underscore the value of CYP51′s key binding sites in guiding virtual screening.

This study integrates dual computational approaches: molecular docking against C. albicans CYP51, a fungal-specific ergosterol biosynthesis enzyme with low homology to human proteins [17], and ligand-based pharmacophore modeling to identify structural motifs critical for antifungal activity. Using these methods, we screened the TCMSP database and identified α-hederin and elemenin as promising candidates. Subsequent validation included in vitro assays (antifungal activity, cytotoxicity, biofilm inhibition) and in vivo evaluation in a murine oral candidiasis model. MD simulations further elucidated α-hederin’s superior binding stability with CYP51 compared to fluconazole. Our findings advance antifungal discovery by bridging computational prediction with mechanistic validation, offering a template for rapid, target-driven development of natural antifungals.

Materials and methods

Strains, cells and growth conditions

The Candida albicans strain CMCC(F)98001 was purchased from Solarbio Biotechnology Co., Ltd (Beijing, China). The strain was revived on YEPD agar medium (1 % yeast extract, 2 % peptone, 2 % glucose and 2 % agar) and subsequently cultured overnight in YEPD liquid medium at 30 °C, with 220 rpm shaking. The 293 T, Raw264.7 cell line were maintained in our laboratory, while the KB (human oral epidermoid carcinoma cells) cell line was obtained from Wuhan Pricella Biotechnology Co., Ltd. Cells were cultured in complete medium containing DMEM (supplemented with glucose, amino acids, vitamins, and trace elements), 10 % fetal bovine serum (FBS), 1 % penicillin, 1 % streptomycin. The DMEM medium was purchased from Biosharp Co., Ltd.

Virtual screening of antifungal natural small molecules targeting CYP51

The crystal structure of C. albicans CYP51 protein (PDB ID:5TZ1) was obtained from the RCSB PDB database. Structural optimization, including hydrogenation and charge assignment was performed using AutoDock tools. Based on literature evidence and active site analysis, the ligand-binding pockets were defined by three critical residues: Tyr118, His310, and Ser378 [20]. Natural small molecules from the TCMSP database were pre-screened using bioavailability (OB > 30 %) and drug-likeness (DL > 18 %) [21]. Virtual screening was subsequently conducted with AutoDock vina, which ranks protein–ligand binding modes by calculating binding free energies derive from the scoring function and molecular force field parameters of C. albicans CYP51 [22].

To enhance screening specificity, a pharmacophore model was developed using the following workflow: (1) CYP51-targeting drugs were queried from the Binding Database to compile a training set; (2) common pharmacophore features of active compounds were extracted and validated to construct the model. For mechanistic validation, the interaction mode of fluconazole with CYP51 was analyzed as a reference. Candidate compounds exhibiting similar binding patterns to fluconazole in the active pocket were prioritized. Hydrogen-bond interactions between these compounds and CYP51′s active residues (Tyr118, His310, Ser378) were further evaluated using Discovery Studio 2019.

Seventeen candidate compounds were experimentally tested for antifungal activity against C. albicans at 12 h, 24 h, and 36 h. A threshold concentration of 1024 μg/mL was established, with compounds failing to inhibit fungal growth at this concentration excluded from further analysis.

Determination of the minimum inhibitory concentration (MICs) of α-hederin and elemicin against C. albicans

The MICs of α-hederin and elemicin against C. albicans were determined using the microbroth dilution method [23]. Briefly, C. albicans cell suspensions (1 × 10⁶ CFU/mL) in YEPD medium were exposed to serially diluted concentrations of α-hederin (0.5–256 μg/mL) or elemicin (0.5–256 μL/mL) and incubated at 37 °C for 24 h and 36 h. Fungal growth was determined by measuring optical density (OD600) using a microplate reader. The MIC was defined as the lowest compound concentration showing ≤ 100 % growth relative to the untreated control (OD600 ≤ control baseline), fluconazole served as the positive control. All experiments were performed in triplicate.

Cytotoxicity assessment of α-hederin and elemicin in Raw264.7, 293 T and KB cells

The cytotoxic effects of the α-hederin and elemicin were evaluated in Raw264.7, 293 T and KB cell lines using the CCK8 assay [24]. Briefly, cells were thawed, cultured, and seeded into 96-well plates at a density of 1 × 10⁵ cells/well. After 24 h of incubation to ensure adhesion, the culture medium was replaced with fresh DMEM containing α-hederin or elemicin at final concentrations ranging from 8 to 128 μg(μL)/mL. The negative control contained cells and DMEM alone, while the blank control consisted of DMEM without cells. Following a 24 h treatment, CCK-8 reagent (10 % v/v) was added to each well, and plates were incubated for 2 h at 37 °C under 5 % CO₂. Absorbance at 450 nm was measured using a microplate reader. Fluconazole served as the reference control, and all experiments were performed in triplicate.

Observation of microstructure changes of C. albicans by SEM and TEM

Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) were employed to evaluated ultrastructural changes in C. albicans following treatment with α-hederin and fluconazole, with untreated cells serving as the control. Sample preparation protocols, including fixation and dehydration were carried out based on the literature [25]. Briefly, C. albicans suspensions before and after treatment with α-hederin and fluconazole were centrifuged, washed three times with PBS, and fixed overnight in 4 % glutaraldehyde. After washing and sequential ethanol dehydration (30 %, 50 %, 75 %, 90 %, 95 % and 100 %, V/V), the samples were critical point dried, mounted on conductive adhesive-coated stubs, and putter-coated with gold. SEM images were taken on a NOVA450 SEM (FEI, Eindhoven, The Netherlands). Similarly, TEM micrographs were taken by centrifuging the cells to collect the precipitate, washing with PBS, and then fixing with 4 % glutaraldehyde solution overnight. The precipitates were then pre-embedded using agar, and the samples were fixed in 1 % osmium acid at room temperature for 2 h in the dark. Then the samples were serially dehydrated in ethanol and permeabilized with acetone and embedding agent. Finally, the samples were sprayed onto a copper grid covered with a holey carbon film for TEM imaging.

Determination of total biofilm biomasses

Crystal violet (CV) is an alkaline dye that simultaneously stains both living and dead cells; thus, it is often used to measure the total amount of cells in biofilm [26]. Firstly, 500 μL of fungal cell suspension (1 × 10⁷ CFU/mL) and 500 μL of RPMI-1640 medium were added to 24-well plate. Then, 200 μL of different concentrations of α-hederin and fluconazole ranging from 16 μg/mL to 256 μg/mL were added to each well and incubated at 37 °C for 24 h. In the control wells, the drugs were replaced with distilled water. After incubation, the supernatant was removed, and the biofilms attached to the plate were gently rinsed three times with sterile phosphate-buffered saline (PBS, pH 6.8), 100 μL of methanol was added to the well to fix the biofilm for 15 min; then, 200 μL of CV solution (0.1 %) was added and maintained for 20 min. The CV solution was removed, and each well was gently washed with distilled water multiple times. Then, 33 % acetic acid was added. The absorbance value was read at 570 nm using a microplate reader. The inhibition rate of biofilm biomass was calculated using the following formula: Inhibition (%) = (OD₅₇₀ blank – OD₅₇₀ treatment)/ OD₅₇₀ blank × 100 %.

Effect of α-hederin on metabolic activity of C. albicans cells during biofilm formation and in mature biofilms

The cell metabolic activity of C. albicans during biofilm formation and in mature biofilm inhibition assays were performed in 96-well microtiter plates using a previously described method with slight modification [27,28]. Briefly, C. albicans suspension in RPMI-1640 medium was prepared at a concentration of 1 × 10⁵ cells/mL and dispensed into 96-well plates (180 μL/well). Serial twofold dilutions of α-hederin and fluconazole (8–128 μg/mL) in RPMI-1640 medium (20 μL/well) was added to disignated wells, while control wells contained RPMI-1640 medium alone. Plates were incubated at 37 °C for 24 h. Biofilm metabolic activity was quantified via the XTT reduction. Absorbance at 490 nm (OD₄₉₀) was measured using a microplate reader, with biofilm viability calculated with the following formula: Viability (%) = (OD₄₉₀ treatment – OD₄₉₀ blank)/ (OD₄₉₀ negative control – OD₄₉₀ blank) × 100 %.

Mature biofilms were established by incubating C. albicans suspensions in RPMI-1640 medium at 37 °C for 24 h. Drug treatments and viability assessments followed the biofilm formation protocol. All experiments were performed in triplicate.

Effect of α-hederin on the hyphal growth of C. albicans

The ability of C. albicans to form hypha was conducted according to the previously reported method [29]. Briefly, RPMI-1640 medium (supplemented with 10 % FBS) was utilized to investigate the characteristics of C. albicans hyphal formation. A suspension of C. albicans at a concentration of 1 × 10⁵ CFU/mL was transferred into 24-well plates. Different concentrations (100 μL) of α-hederin solutions were prepared via gradient dilution. The samples were cultured at 37 °C with an agitation rate of 100 rpm for 24 h. After the experiment, the samples were dehydrated and observed and analyzed using a FEI Nova 450 SEM (FEI, Eindhoven, The Netherlands).

Effect of α-hederin on the cell surface hydrophobicity of C. albicans

The cell surface hydrophobicity was conducted following the methods described by Han with appropriate modifications [29]. First, a cell suspension of C. albicans at 1 × 10⁵ CFU/mL was prepared and placed in PBS. An aliquot of 3 mL of cell suspension were mixed with varying concentrations of α-hederin, then distributed into five borosilicate glass test tubes (18 × 75 mm²). These were cultured in a constant temperature shaker at 28 ℃ for 1 h. Then, 100 μL of the cell suspension from each tube was transferred to 96-well plates for analysis. The initial optical density (A₀) was measured at 600 nm using a microplate reader. Next, for each group of cell suspensions, 1 mL of n-octane was added and incubated at room temperature for 30 min. The aqueous phase was separated by centrifugation and transferred to the corresponding wells on the microplate. Then, another 100 μL sample solution was added to each group. This process was repeated to obtain the final optical density value (A₁). A cell-free control group served as the negative control. Hydrophobicity percentage (CSH%) was calculated using following formula: CSH(%) = (A₀ − A₁) / A₀ × 100 %. To ensure reliability, this entire procedure independently repeated three times.

Determination of ergosterol content in C. albicans

The determination of ergosterol content followed the experimental design of by Yang et.al.[30] and Palacios DS et.al. [31]. Strains in the stationary phase were selected and cultured in a constant temperature shaker at 30 °C for 24–36 h. The samples were then collected after centrifugation (8000 rpm for 10 min), washed with sterile distilled water to remove surface impurities, and freeze-dried. 0.2g of the mycelium powder was placed in a round-bottom flask, and 3 mL of pre-prepared saponification solution (90 % ethanol solution containing 15 % NaOH) was added, followed by 2 mL PBS. The mixture was heated in a water bath at 80 °C for 1 h. After three extractions with petroleum ether, the total volume was adjusted to approximately 20 mL. The supernatant was transferred to another container and subjected to vacuum concentration and dehydration. 1 mL of the extract was taken and the solvent was removed using a rotary evaporator. Then, it was redissolved in 1 mL of anhydrous ethanol. The absorbance values were measured in the range of 230 nm to 281 nm using a UV spectrophotometer. The sterol content percentage (based on wet weight) was calculated using the formula: (OD₂₈₁/OD₂₉₀) × diluted multiple (OD₂₃₀/OD₅₁₈)/100 × 20/0.2. The entire process was repeated three times.

Determination of synergistic antifungal effect of α-hederin and fluconazole against C. albicans

The synergistic antifungal interaction between α-hederin and fluconazole was evaluated using the fractional inhibitory concentration index (FICI). Follwing the checkerboard dilution method described by Segatore [32], serially dilutions of α-hederin (8–128 μg/mL) were prepared in rows A-E of a 96-well plate, while fluconazole concentrations (0.5–8 μg/mL) were serially diluted in columns 1–5. The optical density (OD600) of each well was measured using a microplate reader. The MICs of α-hederin (MIC_Hed) and fluconazole (MIC_Flu) were defined as the lowest drug combination resulting in an OD600 value ≤ 100 % of the growth control.‌ ‌Each experiment was performed in triplicate.‌ The FICI value, calculated using formula (1), was employed to determine the synergistic effect between the two molecules:

$$\mathit{FICI}=\frac{{\mathrm{M}\mathrm{I}\mathrm{C}}_{\mathrm{F}\mathrm{l}\mathrm{u}\mathrm{a}\mathrm{s}\mathrm{s}\mathrm{o}\mathrm{c}\mathrm{i}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}}{{\mathrm{M}\mathrm{I}\mathrm{C}}_{\mathrm{F}\mathrm{l}\mathrm{u}}}+\frac{{\mathrm{M}\mathrm{I}\mathrm{C}}_{\mathrm{H}\mathrm{e}\mathrm{d}\mathrm{a}\mathrm{s}\mathrm{s}\mathrm{o}\mathrm{c}\mathrm{i}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}}{{\mathrm{M}\mathrm{I}\mathrm{C}}_{\mathrm{H}\mathrm{e}\mathrm{d}}}$$ (1)

A synergistic interaction is observed when FICI ≤ 0.5, and the smaller the FICI value, the stronger of the synergistic effect; an additive effect is defined for 0.5 < FICI ≤ 1; no interaction is denoted for 1 < FICI ≤ 2; and antagonism occurs when FICI > 2.

Therapeutic effect of α-hederin on mice oral infected with C. albicans

Male BALB/c mice within the weight range of 25 to 30 g (totaling 25 mice) were selected for the study and provided by the Health Science Center (HSC) of Xi’an Jiaotong University. During the experiment, the mice were housed in single cages with five mice per cage, and were provided with free access to water and standard feed. The room temperature was maintained at 22 ± 2°C, and the light cycle was set at 12-hour alternation of day and night. All mice adapted for one week prior to the experiment to ensure stable physiological conditions. Preliminary statistical analysis indicated no significant difference in initial body weight among groups (P > 0.05). The mice were randomly assigned to five experimental groups (Fig. 7A)[33]. Subsequently, an oral candidiasis model was established by subcutaneous injection of cortisone acetate every other day. C. albicans suspension was prepared at a concentration of 1 × 10⁸ CFU/mL. 1 mL was placed in a 10 mL centrifuge tube and soaked in a sterile cotton swab for about 5 min before using. In this experiment, the mice were first anesthetized (using 5 % chloral hydrate injection at a dose of 10 mL/kg), and then the C. albicans suspension was administered orally, followed by gentle wiping of the tongue surface with a moistened cotton swab for approximately 75 min. During the drug treatment stage, fluconazole was used as the positive control group due to its significant efficacy against C. albicans infection. In the experiment, α-hederin was added to the drinking water at concentrations of 32 μg/mL (labeled as 32Hed) and 128 μg/mL (labeled as 128Hed) for five days; another group was given 128 μg/mL of fluconazole daily. The treatment was initiated 1 day following C. albicans infection and continued for 5 days.

Fig. 7.

Molecular dynamics simulation of the interaction between α-hederin and fluconazole with CYP51. (A) RMSD plot of α-hederin and fluconazole with CYP51. (B) RMSF plot of α-hederin and fluconazole with CYP51. (C) Amino acid free energy contribution of the composite system of α-hederin (D) and fluconazole (E) with CYP51. Free energy decomposition of α-hederin and fluconazole with CYP51, respectively. (F) Rg of the simulated process of α-hederin and fluconazole. (G) SASA of α-hederin (H) and fluconazole (I) simulation processes. the number of hydrogen bonds in α-hederin and fluconazole with CYP51 complex system, respectively.

Measurements of animal parameters

The changes in the body weight of mice were recorded throughout. After the experiment, general anesthesia was administered using ether. Venous blood around the eyes was collected and stored in sterile tubes. Meanwhile, whole blood was collected in tubes with EDTA (ethylenediaminetetraacetic acid) anticoagulant for complete blood count analysis. These data were obtained by Sysmax XN-550 hematology analyzer (Shanghai, China). The isolated serum samples were aliquot into an ultra-low temperature refrigerator at −80 °C for subsequent testing. To determine the degree of C. albicans infection, the tongues were bisected and extracted post-mortem under sterile conditions and then ground evenly. A 100 μL sample of the homogenate was taken and placed in the chromogenic medium, and incubated at 28 °C for 48 h. Subsequently, the counting method was used to count the colony-forming units (CFU) and the results were expressed by the CFU/g tongue. After the remaining tissue samples were treated with PAS staining, the relevant situations were observed and recorded using the PANNORAMIC 250 FLASH slide scanner (3DHISTECH, Hungary).

Molecular dynamics simulation of the interaction between α-hederin and CYP51

Molecular dynamics simulations were conducted using Gromacs 4.5.1 software. The protocol followed the procedure outlined by Yu [34]. Specifically, the structure files of α-hederin, fluconazole, and CYP51 were processed to generate a.top file as an input for Gromacs. An appropriate simulation box size was selected based on the dimensions of the composite system. Energy minimization was performed using the steepest descent method followed by the conjugate gradient method. Subsequently, solutes were restrained for restricted simulations, and the system was gradually heated to 300 K. An NPT ensemble simulation was then initiated to adjust the system pressure until density convergence was achieved. Once the system reached stability, as evidenced by converged pressure and density, data collection commenced. The binding free energy of the α-hederin and fluconazole complexes with CYP51 was calculated during the 100 ns simulation phase. Additionally, the free energy contributions of individual amino acids were analyzed, and the binding free energy was decomposed into van der Waals (vdw), electrostatic (ele), polar solvent, and nonpolar solvent components to elucidate the binding mechanisms of α-hederin and fluconazole. Kinematic parameters of the trajectories of each atom in the α-hederin and fluconazole complexes with CYP51 were calculated during the 100 ns simulation phase. Root-mean-square deviation (RMSD) values were used to assess the average deviation of the complex conformation from its initial state over time, thereby determining whether the system had reached equilibrium. Root-mean-square fluctuation (RMSF) values were computed to quantify the flexibility of each residue, providing insights into localized protein mobility [35]. The RMSD and RMSF curves were plotted against the number of residues to analyze their morphological and amplitude characteristics along the kinetic simulation trajectory.

The radius of gyration (Rg) was calculated for the α-hederin and fluconazole complexes with CYP51 during the 100 ns simulation phase using the following formula: Rg=${\sqrt{\frac{1}{n}}{\sum }_{i=1}^n(r_i-r_{\mathit{cm}})}^2$,

in which Rg denotes the radius of gyration; r_i denotes the position vector of the i-th atom or unit; and r_cm is the position vector of the center of mass.

In molecular dynamics simulations, hydrogen bonds were identified based on geometric and energetic criteria. Geometric criteria included distance and angle thresholds, while interaction energies between donor, hydrogen, and acceptor atoms were calculated to confirm hydrogen bond formation. If the interaction energy fell below a predefined threshold, the hydrogen bond was deemed formed [36]. Within the 100 ns simulation phase, interatomic distances and interaction energies between α-hederin, fluconazole, and the CYP51 complex were evaluated to identify all possible donors and acceptors. Time-series data were generated to monitor changes in hydrogen bond numbers throughout the simulation.

Results

Virtual screening of antifungal natural small molecules targeting CYP51

The high resolution C. albicans CYP51 protein structure (PDB: 5TZ1 Fig. 1A) was selected and subjected to conformation optimization (Fig. 1B). Key active site amino acids (Tyr118, His310, and Ser378) were identified, and an active pocket was constructed within the β5 region (Fig. 1C), positioned above the heme bulge. Notably, the β5 region is unique to fungal CYP51 isoforms, making it an ideal target for screening natural compounds with fungal-specific activity while minimizing off-target effects in humans.

Fig. 1.

Crystal structure (A), and the lowest energy conformation of 5TZ1(B) and comparation of the CYP51 β5 sequences in C. albicans and human(C).

The virtual screening workflow is illustrated in Fig. 2. First, the TCMSP database was prefiltered using oral bioavailability (OB) and drug-likeness (DL) criteria. Next, the LibDock scoring function in Discovery Studio was employed to predict ligand conformations, followed by energy minimization using the CDOCKER module. Subsequently, binding energies were calculated with AutoDock vina. Natural small molecules identified through this molecular docking approach (25 compounds) were cross-referenced with those screened via Discovery Studio’s pharmacophore model (57 compounds). Compounds demonstrating strong hydrogen bonding interactions with CYP51 at the active site were prioritized. The intersection of these two screening methods yielded 17 candidate molecules (Fig. 2). The antifungal activity of these 17 compounds against C. albicans was evaluated (Table 1). Two natural small molecules-α-hederin and elemicin-exhibited potent antifungal activity at concentrations below 128 μg/mL, confirming their potential as effective antifungal agents.

Fig. 2.

Workflow of virtual screening target to CYP51.

Table 1.

Antifungal activity of candidate compounds.

Molecule ID	Molecule name	Anti − C. albicans activity
MOL001951	Bergapten	>1024 μg/mL
MOL002395	Deoxyandrographolide	>1024 μg/mL
MOL003958	Evodiamine	>1024 μg/mL
MOL005236	Gibberellin A3	>1024 μg/mL
MOL005329	Ginsenoside Ra2	>1024 μg/mL
MOL005331	Ginsenoside Rb1	>1024 μg/mL
MOL005344	Ginsenoside Rh2	>1024 μg/mL
MOL003225	Hypodiolide A	>1024 μg/mL
MOL003617	Isogosferol	>1024 μg/mL
MOL003217	Isoxanthohumol	>1024 μg/mL
MOL002809	Picroside Ⅱ	>1024 μg/mL
MOL000040	Scopoletol	>1024 μg/mL
MOL013218	Strictosamide	>1024 μg/mL
MOL004561	Sudam Ⅲ	>1024 μg/mL
MOL003232	Triptofordin	>1024 μg/mL
MOL000269	Elemicin	<128 μL/mL
MOL010359	α-Hederin	<128 μg/mL
Reference drug	Fluconazole	>128 μg/mL

Minimum inhibitory concentration and cytotoxicity of candidate compounds against C. albicans

The minimum inhibitory concentrations (MICs) of α-hederin, elemicin and fluconazole against C. albicans are summarized in Fig. 3A-C. Both α-hederin and elemicin demonstrated potent antifungal activity, with MIC values of 32 μg/mL and 16 μL/mL, respectively. Notably, elemicin inhibited 50 % −70 % of C. albicans growth at concentrations exceeding 0.5 μL/mL, while α-hederin achieved 50 % growth inhibition at concentrations above 8 μg/mL within 24 h. In contrast, fluconazole demonstrated markedly inferior antifungal activity compared to α-hederin and elemicin, as evidenced by persistent fungal growth even at 128 μ g/mL after 24 h.

The cytotoxicity of α- hederin, elemicin and fluconazole were assessed by incubating 293 T, Raw264.7 and KB cells with vary concentrations of each compound for 24 h. Cell viability results are shown in Fig. 3D-F. Both α-hederin and fluconazole exhibited low cytotoxicity, even at the highest tested concentration (128 μg/mL), cell survival rates remained above 80 % across all three cell lines. In contrast, elemicin demonstrated significant toxicity to 293 T, Raw264.7, and KB cells at concentrations higher than 8 μL/mL (Fig. 3E). The 90 % cytotoxic concentration (CC90) was determined to be 93.15 μg/mL for α-hederin, 60.65 μg/mL for fluconazole, and 0.72 μL/mL for elemicin. Based on its favorable safety profile, α-hederin was selected for further investigation of its antifungal mechanisms in subsequent studies.

Fig. 3.

Minimum inhibitory concentrations of α-hederin (A), elemicin (B) and fluconazole (C) against C. albicans, and their cytotoxicity to Raw264.7 (D), KB (E), and 293 T cells (F). The statistical differences between groups (≥3) were evaluated with two-way analysis of variance using the Tukey test. (The same letter indicates no significant difference (P > 0.05), adjacent letters indicate significant difference (P < 0.05), and alternating letters indicate extremely significant difference (P < 0.01)).

Effect of α-hederin on the microstructure of C. albicans

The structural alterations in C. albicans treated with α-hederin was observed using scanning electron microscopy (SEM) and transmission electron microscopy (TEM), as shown in Fig. 4A. α-Hederin-treated C. albicans exhibited severe surface deformation, including cell concavity and pronounced roughness. In contrast, fluconazole-treated cells retained a smooth, intact surface with no structural abnormalities, highlighting α-hederin’s superior ability to disrupt fungal cell integrity. The internal microstructure of C. albicans was observed by TEM, and it was found that untreated control cells (CK) showed a dense cytoplasm rich in electron-dense particles, well-defined organelles, intact nuclear membranes, and smooth cell walls/membranes. In stark contrast, α-hederin treatment (32 μg/mL) caused drastic intracellular changes including severe cytoplasmic content loss and vacuolization, thinning and irregularity of cell walls/membranes and membrane discontinuities, surface protrusions, and near-empty cellular compartments. These findings suggest that α-hederin exerts antifungal activity primarily by destabilizing membrane integrity, leading to leakage of cellular contents. Its structural disruption was markedly more pronounced than that of fluconazole, which showed no comparable ultrastructural damage.

Fig. 4.

Effects of α-hederin and fluconazole on the micromorphological structure and biofilm formation and mature biofilms. (A) Microstructure of C. albicans observed by SEM and TEM, (B) The C. albicans biofilms biomass detected by crystal violet staining method, (C, D) Analysis results of total biofilm biomass by crystal violet staining. (E-H) The metabolic activity of C. albicans cell during the biofilm formation (E, F) and in mature biofilms (G, H) detected by XTT method. (The same letter indicates no significant difference (P > 0.05), adjacent letters indicate significant difference (P < 0.05), and alternating letters indicate extremely significant difference (P < 0.01)).

Effects of α-hederin on biofilm formation and mature biofilm of C. albicans

The formation of biofilms is an important pathogenic factor of C. albicans. The inhibitory effects of α-hederin on biofilm formation and mature biofilms were investigated using the XTT assay to assess the metabolic activity of cells within biofilms and crystal violet staining to quantify total biofilm biomass. As shown in Fig. 4B-F, α-hederin inhibited early biofilm formation of C. albicans in a concentration-dependent manner, achieving 55 % and 75 % inhibition at 32 and 256 μg/mL, respectively. In contrast, fluconazole was significantly less effective, showing only 57 % inhibition even at 256 μg/mL. This suggests α-hederin is more potent at preventing biofilm formation, potentially by impairing fungal adhesion. However, neither agent effectively eradicated mature biofilms (Figs. 4G, H), with inhibition rates of only 20 % and 17 % for α-hederin and fluconazole, respectively, at 128 μg/mL. Collectively, these data indicate that α-hederin exerts its most potent anti-biofilm effects during the early formation stage, highlighting its promise for targeting biofilm-related infections.

Effects of α-hederin on hyphal formation, cell surface hydrophobicity, and ergosterol content in C. albicans

Hyphal formation is critical for C. albicans virulence, as it enhances host cell adhesion and tissue invasion. α-Hederin demonstrated potent inhibition of hyphal development. As shown in Fig. 5A, α-hederin could completely inhibit the hyphal formation of C. albicans at a concentration of 256 μg/mL, with cells exhibiting surface adhesion, dissolution, and aggregation. At 128 μg/mL, partial inhibition occurred, resulting in wrinkled cell surfaces and residual hyphal growth. In contrast, fluconazole only reduced hyphal formation at concentrations > 32 μg/mL, and failed to achieve complete suppression even at ‌256 μg/mL. For cell surface hydrophobicity (CSH), it correlates strongly with C. albicans drug resistance. As shown in Fig. 5 (B, C), α-hederin reduced CSH by 85 % at 128 μg/mL in a concentration-dependent manner. Fluconazole, however, showed minimal efficacy, achieving only 20 % CSH reduction‌ at the same concentration.

Fig. 5.

Effects of α-hederin and fluconazole on pathogenic factors of C. albicans and their combined effects. (A) Hypha formation, (B, C) Cell surface hydrophobicity, (D) Ergosterol synthesis, (E) Synergistic antifungal effects of fluconazole and α-hederin on C. albicans, the color scale represents the cell viability (OD value) under different concentration combinations, the lighter the color, the lower of OD value and the stronger of antifungal effect. (The same letter indicates no significant difference (P > 0.05), adjacent letters indicate significant difference (P < 0.05), and alternating letters indicate extremely significant difference (P < 0.01).

Meanwhile, given α-hederin’s proposed mechanism as a CYP51 inhibitor-a key enzyme in ergosterol biosynthesis-we determined its impact on ergosterol content in C. albicans cells, as shown in Fig. 5D. At ‌16 μL/mL, α-hederin reduced ergosterol content by ‌90%‌ compared to the untreated control (CK), whereas fluconazole achieved only a ‌45% reduction. These results confirm that α-hederin exhibits significantly stronger inhibitory activity against ergosterol synthesis than fluconazole, consistent with its direct targeting of CYP51.

Synergistic antifungal effects of α-hederin and fluconazole on C. albicans

Combination therapy with antifungal agents can mitigate resistance associated with monotherapy. We evaluated the interaction between α-hederin and fluconazole using the fractional inhibitory concentration index (FICI), as shown in the formula (1). As shown in Fig. 5E, the FICI average value was 0.5078 and the lowest observed FICI value of 0.2578. This FICI value indicates an additive antifungal effect of the combination. Fluconazole alone failed to fully inhibit C. albicans growth even at its maximum test concentration (128 μg/mL), suggesting its true MIC exceeds this threshold. Despite using 128 μg/mL as a conservative proxy for MIC_Flu, the combination still demonstrated additive effect at significantly lower concentration (1 μg/mL fluconazole and 16 μg/mL α-hederin). These findings highlight the potential of α-hederin to enhance fluconazole’s antifungal activity while reducing therapeutic dosages.

Therapeutic effects of α-hederin on mice orally infected with C. albicans

An oral C. albicans infection model was established using BALB/c mice. The infected mice were treated with varing concentrations of α-hederin, with fluconazole serving as the control. Changes in body weight during the treatment were monitored as shown in Fig. 6C. From the day of modeling until euthanasia, the normal control group (NC) exhibited a steady increase in body weight, while the susceptible group (Can) showed continuous weight loss post-infection. However, starting from the second day post-modeling, mice treated with 128 μg/mL α-hederin (128 Hed), 32 μg/mL α-hederin (32 Hed), and 128 μg/mL fluconazole (128 Flu) displayed a rebound in body weight. Notably, the 128 Hed group showed the slowest initial weight loss, followed by a consistent increase over the treatment period, indicating that all the therapeutic interventions significantly alleviated C. albicans induced weight reduction.

Fig. 6.

Therapeutic effect of α-hederin on oral infection of C. albicans in mice. (A) Time points and treatments of oral Candidiasis model in the study. Plate coating (B) and colony count (D) of the amount of C. albicans loaded on the tongue of each group of mice. (C) Changes in body weight of mice infected with C. albicans after α-hederin and fluconazole treatment. (E, F, G) Hematological characteristics of leukocyte (WBC), hemoglobin (HGB) and platelet (PLT) in mice, respectively. (H) Morphology and characteristics of mouse tongue after α-hederin and fluconazole treatment and PAS staining of tongue tissue sections. Each group consisted of five mice. Red circles and arrows show the growth of C. albicans; green arrows show mycelium-like C. albicans.Abbreviations: NC, noninfected mice; Can, infected, but not treated; 128Hed, infected, treated with 128 μg/mL per day α-hederin;32Hed, infected, treated with 32 μg/mL per day α-hederin; 128Flu, infected, treated with 128 μg/mL per day fluconazole.

Fungal colonization on the tongues of infected mice was detected using periodic acid-Schiff (PAS) staining (Fig. 6H) and quantified using agar plate culture (Fig. 6B). PAS staining revealed no C. albicans attachment in the CK group, which exhibited a normal pink tongue with a smooth, intact surface. In contrast, the Can group showed extensive fungal adhesion and tissue damage. α-Hederin treatment markedly reduced fungal colonization and mitigated tongue surface injury. Colony counts confirmed a significant decrease in fungal load in α-hederin-treated group (Fig. 6D). Remarkably, no colonies were detected in the 128Hed group, indicating complete suppression of oral C. albicans infection at this concentration.

Key hematological indices are shown in Fig. 6(E, F, G). Compared to the CK group, the Can group showed reduced white blood cell count (WBC), and hemoglobin (HGB) level, alongside elevated platelet (PLT) counts, suggesting immunosuppression post infection. In contrast, all treatment groups (128 Hed, 32 Hed, and 128 Flu) showed restored WBC and HGB levels and normalized PLT counts relative to the Can group. These results indicate that both α-hederin and fluconazole effectively counteracted C. albicans-induced toxicity and partically restored immune function in mice.

Molecular dynamics mechanism of the interaction between α-hederin and CYP51

Molecular dynamics simulations of the α-hederin-CYP51 and fluconazole-CYP51 complex systems were performed for 100 ns, and the results are summarized in Fig. 7. The root means square deviation (RMSD) analysis (Fig. 7A) revealed stabilization of the α-hederin-CYP51 and fluconazole-CYP51 systems at 0.18 nm and 0.23 nm, respectively, indicating greater structural stability in the α-hederin complex. Root mean square fluctuation (RMSF) values (Fig. 7B) showed increased flexibility in the 70–100 and 180–200 residue regions of CYP51 upon α-hederin binding. Binding free energies calculations (Fig. 7C) further showed that residues in the 50–150, 170–200, and 400–450 regions of the α-hederin complex exhibited lower energy values compared to fluconazole, suggesting enhanced stabilization of these regions in the α-hederin-CYP51 system. Consistent with this, the α-hederin complex displayed significantly lower overall binding free energy than the fluconazole complex (Fig. 7, Fig. 7).

Structural compactness was assessed via radius of gyration (Rg) analysis (Fig. 7F). Both systems stabilized at about 2.3 nm after 40 ns, reflecting similar structural densities and stability during simulation. Solvent-accessible surface area (SASA) values (Fig. 7G) fluctuated within 220–230 nm² for both complexes over 100 ns, indicating comparable structural compactness of CYP51 in both systems. Notably, the α-hederin complex formed an average of 4–5 hydrogen bonds with CYP51, whereas the fluconazole complex maintained only 1–2 (Fig. 7, Fig. 7). This stronger hydrogen-bond network, combined with additional stabilizing interactions such as alkyl interactions, carbon-hydrogen bonding, and other kinds of interactions (Fig. 8B), explains the superior stability and lower energy state of the α-hederin-CYP51 complex compared to fluconazole (Fig. 8A).

Fig. 8.

Interaction and bonds formed in fluconazole-CYP51 complex (A) and α-hederin-CYP51 complex (B).

Discussion

In recent years, computer-aided drug discovery (CADD) techniques have greatly accelerated the pace of drug development. In particular, the emergence of methods such as virtual screening and deep learning has revolutionized antimicrobial drug discovery by enabling rapid identification of candidate molecules [37,38]. Current CADD approaches for antimicrobial discovery primarily focus on either target protein-based screening or ligand property-driven design using pharmacophore modeling and deep learning. However, challenges such as inaccurate intramolecular energy calculations and inherent limitations of molecular docking often led to high false-positive rates, hindering practical applications[39,40].

To address these limitations, we propose an innovative dual-target strategy that integrates both protein and ligand screening parameters. By intersecting results from these complementary approaches, we reduced the candidate pool from over 1,000 compounds to 25 high-probability targets, dramatically improving screening efficiency and accuracy. This target-centric approach addresses the common challenge of unknown antifungal targets in virtual screening. Applying this strategy to antifungal discovery, we screened natural small molecules targeting CYP51 (the therapeutic target of fluconazole) and identified two novel compounds, α-hederin and elemicin. Notably, these molecules exhibit a mechanism of action similar to fluconazole but demonstrate superior antifungal activity, positioning them as promising candidates for further development.

Hederin, a pentacyclic triterpenoid isolated from ivy leaves, derives its pharmacological activity primarily from saponin constituents. Among these, α-hederin and hederin C exhibit broad therapeutic potential. Hederin C has shown anti-inflammatory and antimicrobial activities, including inhibition of MAPK/NF-κB signaling in human intestinal epithelial Caco-2 cells [41] and mitigation of Staphylococcus aureus-induced inflammation in RAW 264.7 macrophages [42]. α-Hederin, however, is best characterized for its potent antitumor effects. It inhibits cancer cell proliferation and metastasis by inducing apoptosis and ferroptosis, demonstrating efficacy across diverse tumor cells [43,44]. Recent studies suggest additional antifungal mechanisms, including disruption of C. albicans cell membrane integrity and interference with fungal metabolic pathways [45,46].

Our experimental data revealed that α-hederin (32 μg/mL) completely inhibited C. albicans growth, with no fungal recovery observed after 36 h of treatment, whereas the fungi treated with fluconazole regrowth after 36 h, demonstrating its strong antifungal activity. Further mechanistic studies demonstrated its ability to inhibit the hyphal formation and biofilm formation in C. albicans. Moreover, in vivo validation using a murine oral infection model confirmed α-hederin’s superior efficacy over fluconazole in controlling C. albicans infections. Collectively, these findings indicate that α-hederin is a promising candidate for antifungal therapy, offering both enhanced potency and a unique mechanism of action.

Simultaneously, our study revealed promising antifungal activity in elemicin, a novel alkenyl benzene compound identified through virtual screening. This natural small molecule, commonly found in herbs and spices, exhibited inhibitory effects against C. albicans. Pinheiro et al. corroborated elemicin’s antifungal efficacy against Aspergillus niger and Aspergillus flavus when extracted from Calyptranthes concinna leaves [47]. However, cytotoxicity assays of elemicin in our study revealed significant toxicity to 293 T, Raw264.7 and KB cells at concentrations > 8 μL/mL. Genotoxic and carcinogenic properties of elemicin, as reported by Mario et al. [48], further emphasize the need for structural optimization to enhance its therapeutic safety profile.

In parallel, this study elucidated the antifungal mechanism of α-hederin. Using molecular docking and pharmacophore modeling, we identified CYP51 (lanosterol 14α-demethylase) as the antifungal target of α-hederin. It demonstrated superior ergosterol synthesis inhibition compared to fluconazole. As CYP51 is pivotal in ergosterol production, its inhibition directly reduces ergosterol levels [49], validating the virtual screening results.

To further characterize α-hederin’s interaction with CYP51, molecular dynamics simulations were performed. Comparative analyses of α-hederin-CYP51 and fluconazole-CYP51 complexes evaluated trajectory stability (RMSD, RMSF), binding free energy, radius of gyration (Rg), hydrogen bonding patterns, and solvent-accessible surface area (SASA) [50]. The α-hederin complex exhibited enhanced structural stablity and tighter CYP51 compaction, and a stronger hydrogen-bond network, explaining its superior antifungal efficacy at the molecular level.

Given fluconazole’s widespread resistance issues, we investigated α-hederin’s combinatorial potential. Synergistic effects were observed when α-hederin was combined with fluconazole, reducing the required fluconazole dosage while maintaining efficacy. This phenomenon may arise from α-hederin’s dual mechanisms: robust CYP51 binding stability and inhibition of fungal biofilm and hyphae formation. While the precise molecular basis warrants further investigation, these findings suggest α-hederin as a standalone therapeutic or combination adjuvant, potentially mitigating toxicity and resistance associated with conventional antifungals.

Conclusions

This study employed a dual virtual screening strategy combining molecular docking and pharmacophore modeling, targeting CYP51, to systematically screen natural small molecules from the TCMSP database. Through iterative optimization of the screening protocol, 17 candidate compounds were identified from a large pool of natural molecules. Experimental validation further narrowed this selection to two antifungal natural products, α-hederin and elemicin. Notably, α-hederin demonstrated potent antifungal activity against C. albicans with low cytotoxicity, primarily inhibiting critical virulence factors such as hyphal formation and biofilm development. Its therapeutic efficacy was corroborated in vivo using a murine oral candidiasis model, underscoring its potential for clinical translation. Molecular dynamics simulations at the atomic level elucidated the interaction mechanisms of α-hederin and fluconazole with CYP51, revealing α-hederin’s superior binding stability. This work not only identifies α-hederin as a promising novel therapeutic candidate for fungal infections but also establishes a methodological framework for accelerating antifungal drug discovery and mechanistic studies.

Data analysis

In this study, GraphPad Prism 10.0 (GraphPad Software Inc., CA, USA) was used to conduct statistical analysis on the experimental data. Involving key parameters such as in vitro growth inhibition, PAS staining, biofilm maturation and analysis of hematological indicators, one-way ANOVA was conducted to confirm the statistical differences among each group, multiple comparisons were then performed using Dunnett's t-test, and the results were finally presented in the form of mean ± standard deviation. If the P value was less than 0.05, it indicated that the differences between groups were significant.

Compliance with Ethics Requirements

This study was conducted in accordance with the ethical guidelines set forth by the Northwestern Polytechnical University and adhered to the principles outlined in the Declaration of Helsinki. For animal experiments, the research protocol was reviewed and approved by the Northwestern Polytechnical University Medical and Laboratory Animal Ethics Committee with protocol number (202501151). The study was carried out following the recommendations of the Guide for the Care and Use of Laboratory Animals. All efforts were made to minimize suffering and ensure the highest standards of care for animals involved in the study.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.