Algerian honeybee venom antioxidant and antimicrobial activities: molecular docking of melittin as a dual inhibitor of PBP2a and Keap1

Honeybee venom is a complex bioactive secretion dominated by melittin, with recognized antioxidant and antimicrobial therapeutic potential. This study characterized Algerian honeybee venom (AHBV) through biochemical profiling, in vitro antioxidant and antimicrobial assays, and in silico molecular docking to explore melittin’s dual targeting capacity. AHBV was characterized using size-exclusion fast protein liquid chromatography, Fourier-transform infrared spectroscopy, the Bradford assay, the ABTS radical-scavenging assay, antimicrobial evaluation, and molecular docking of melittin against PBP2a and Keap1 using pyDock. Size-exclusion fast protein liquid chromatography revealed a protein profile dominated by low-molecular-weight peptides (<5 kDa), with only minor higher-molecular-weight components, consistent with the known composition of Apis mellifera venom. Fourier-transform infrared spectroscopy confirmed the proteinaceous, peptide-rich nature of the venom through characteristic amide I and II bands. Total protein content, determined by the Bradford assay, was 204.8 ± 16.26 μg BSA equivalents/mL. AHBV exhibited marked antioxidant activity in the ABTS radical-scavenging assay (727.0 ± 4.9 μmol Trolox equivalents/mg). Antimicrobial evaluation demonstrated significant activity against Gram-positive bacteria, particularly Staphylococcus aureus (MIC 15.62–62.5 μg/mL), moderate efficacy against Gram-negative strains (MIC 125–500 μg/mL), and measurable antifungal activity against Candida albicans (MIC 250 μg/mL). Molecular docking of melittin against PBP2a and Keap1 revealed a potential dual targeting: melittin is predicted to bind the PBP2a allosteric pocket with a pyDock score of −40.406 kcal/mol, and potentially engage the Keap1 Kelch domain (−31.537 kcal/mol) through hydrophobic contacts critical for Nrf2 recognition. This interaction suggests a mechanism for resensitizing MRSA to β-lactam antibiotics, indicating a possible role in activating the antioxidant response pathway. Taken together, these findings provide region-specific data on Algerian honeybee venom and position melittin as a promising bifunctional therapeutic candidate against antimicrobial resistance and oxidative stress-related disorders.

Honeybee venom (HBV) serves as a distinctive defense mechanism within the animal kingdom, playing a crucial role in protecting the bee colony. This substance is efficient and possesses a unique biocomplex mixture that demonstrates diverse pharmacological, toxicological, and pleiotropic properties. Researchers have identified at least 18 active components in its composition, including enzymes, peptides, and biogenic amines (Ullah et al., 2023). The primary components of HBV generally include melittin, widely recognized as its major peptide constituent, along with apamin, secapin, procamine, adolapin, and mast cell degranulating peptide (MCDP) (Gajski et al., 2024). Phospholipase A 2 (PLA 2) is the primary enzyme associated with HBV, whereas hyaluronidase, acid phosphomonoesterase, lysophospholipase, and β-glucosidase are present at significantly lower levels (Wehbe et al., 2019). HBV not only contains peptides, but also various non-peptide components that influence numerous cellular systems. These include histamine, dopamine, norepinephrine, sugars like fructose and glucose, and phospholipids (Gajski et al., 2024). HBV has been used in traditional medicine for the treatment of several ailments, including arthritis and dermatological disorders. Besides its anti-inflammatory properties, HBV also has antioxidant, antibacterial, and analgesic actions, indicating its potential as a candidate for enhancing diabetic wound healing (Maitip et al., 2021). A high amount of melittin boosts the antifungal power of HBV, effectively fighting against various types of fungi like Aspergillus, Botrytis, Candida, Fusarium, Malassezia, Neurospora, Penicillium, Saccharomyces, and Trichoderma (Memariani and Memariani, 2020). Its antimicrobial activity is also well documented against both Gram-positive and Gram-negative bacteria, regardless of melittin purification, crude venom composition, or bee species (El-Seedi et al., 2020). Apamin and phospholipase A 2 have potential therapeutic applications in various immune disorders due to their capacity to modulate immune responses (El-Bilawy et al., 2025). HBV is known for its efficacy in human and animal medicine, as well as its potential applications in plant protection and microbial control (El-Bilawy et al., 2025). Due to its broad biological activity, it is crucial to understand the factors that influence its compositional variability to assess its therapeutic potential accurately. Phytogeographical characteristics primarily determine the chemical composition of HBV, which exhibits considerable variability. Comparative studies reveal notable differences in HBV composition across various sources, as determined by chemical and biological assays. Environmental factors, botanical resources, bee genetics, and location influence these differences (Aydeniz-Guneser et al., 2025; Scaccabarozzi et al., 2021).

Based on previous studies of the biological and chemical properties of HBV, this study investigates the composition and associated activities of bee venom collected in Northeastern Algeria. Due to its phytogeographical specificity, the sample was chemically characterized using size-exclusion FPLC to analyze the molecular weight distribution of major protein components, FTIR spectroscopy to determine functional groups and the overall chemical fingerprint, and the Bradford assay to evaluate total protein content. The biological potential of Algerian HBV was evaluated through in vitro antimicrobial assays and the ABTS radical scavenging assay. Furthermore, molecular docking simulations were performed using melittin, selected for its well-established predominance and biological relevance in HBV, as reported in the literature, to explore its potential interactions with PBP2a and Keap1. These computational analyses were intended to generate structural hypotheses regarding possible mechanisms underlying the observed biological activities rather than to provide conclusive evidence of molecular mechanisms.

General chemicals and reagents

All chemicals and reagents were of analytical grade and purchased from Sigma–Aldrich (St. Louis, MO, USA), unless otherwise specified. Ultrapure water was used to prepare all solutions.

Sample collection

Bee venom samples were collected between July and September 2022 from Apis mellifera colonies located in the northeastern region of Algeria (Setif) (Figure 1). Venom was obtained using a non-lethal electrostimulation method optimized for large-scale collection. The collection apparatus consisted of a glass plate fitted with stainless-steel electrodes connected to a power supply set at 1–3 V. The device was positioned at the hive entrance, where a mild electrical stimulus induced a stinging response, resulting in venom deposition on the glass surface. Each collection session lasted approximately 15 min. The dried venom was carefully scraped from the glass using a sterile spatula, transferred into dark, airtight glass containers, and stored at −20 °C until further analysis.

Chemical fingerprinting by FTIR-ATR spectroscopy

Dried HBV powder was analyzed by Fourier-transform infrared spectroscopy (FTIR) coupled with Attenuated Total Reflectance (ATR) to characterize its overall chemical composition, following the method described by Gómez-García et al. (2024) with minor modifications. Infrared spectra were acquired using a PerkinElmer Paragon 1000 FTIR spectrometer (Waltham, MA, USA) with a Diamond/ZnSe ATR accessory. Approximately 30 mg of the HBV powder was evenly spread as a thin layer, and spectra were recorded over 4000–400 cm -1 at 4 cm -1 resolution with 32 scans. Three replicate spectra were collected from different aliquots, and vibrational bands were identified based on the literature.

Analysis by size exclusion chromatography (FPLC)

The protein profile was assessed by gel filtration chromatography under the following conditions: the column was operated at a flow rate of 0.5 mL/min, using phosphate buffer (0.025 M, pH 7) supplemented with NaCl (0.15 M) and NaN3 (0.02% w/v). Protein standards having established molecular weights (Thyroglobulin, 669 kDa; Ferritin, 440 kDa; Aldolase, 158 kDa; Conalbumin, 75 kDa; Ovalbumen, 43 kDa; Carbonic anhydrase, 29 kDa; Ribonuclease A, 13.7 kDa; Aprotinin, 6.5 kDa) were used to construct a calibration curve. The AKTA pure 25 L system, manufactured by GE Healthcare Life Sciences in Freiburg, Germany, was configured with two pressure-controlled pumps for column protection. It included a gel filtration column prepacked with Superdex® 200 10/300 GL, connected in series with a Superdex Peptide 10/300 GL column (also from GE Healthcare Life Sciences, Freiburg, Germany), and a UV multiwavelength detection monitor U9-L, set to a fixed wavelength of 280 nm. The program used for sample evaluation was UNICORN 7.0.

Protein content determination

The soluble protein concentration was quantified using the (Bradford, 1976) assay with minor modifications, as detailed by Gómez-García et al. (2021). In this procedure, 0.950 mL of Bradford reagent was mixed with 0.05 mL of AHBV at room temperature (20 °C ± 0.5 °C). The mixture was vortex-mixed and subsequently maintained in the dark for 20 min. Absorbance was measured at 595 nm (Synergy H1; BioTek Instruments, Winooski, VT, USA), and protein content was determined using a calibration curve with bovine serum albumin (BSA) as the standard (25–300 μg/mL). Results were quantified as mg BSA/mL. All measurements were conducted in triplicate.

Antioxidant potential determination

The ABTS radical scavenging assay was used to evaluate the antioxidant capacity of the samples based on their ability to neutralize the ABTS•+ radical cation, as described by Coscueta et al. (2020). The ABTS•+ stock solution was prepared by mixing ABTS (7 mM) with potassium persulfate (2.5 mM) in ultrapure water and allowing the mixture to react for 16 h at room temperature in the dark. The resulting solution was then diluted with ultrapure water to obtain an absorbance of 0.700 ± 0.020 at 734 nm using a Synergy H1 microplate reader (Vermont, USA).

In a 96-well microplate, 20 µL of sample extract was mixed with 180 µL of ABTS•+ solution and incubated for 5 min in the dark. The absorbance was measured at 734 nm. Ultrapure water was used as blank. Trolox was used as a standard antioxidant, with calibration curves prepared using concentrations ranging from 25 to 175 µM.

Antimicrobial activity

Antimicrobial activity was evaluated against reference and clinical strains obtained from the Laboratory of Applied Microbiology at the University of Setif 1. The bacterial strains included Escherichia coli (ATCC 25922), Staphylococcus aureus (ATCC 6538P, ATCC 25923), Pseudomonas aeruginosa (ATCC 27853), Klebsiella pneumoniae (ATCC 13883), Salmonella enterica (ATCC 13076), Escherichia coli (ATCC 8739), and Bacillus subtilis (ATCC 6633). Antifungal activity was evaluated against Candida albicans (ATCC 10231). Minimum inhibitory concentration (MIC) and minimum bactericidal/fungicidal concentration (MBC/MFC) were determined by broth microdilution according to Mouffok et al. (2023), with minor modifications. Two-fold serial dilutions of bee venom (3.90–500 μg/mL) were prepared in 2× Mueller–Hinton broth in 96-well plates. Each well was inoculated with 20 µL of microbial suspension adjusted to 0.5 McFarland standard. After incubation, 20 µL of 0.01% resazurin was added, and plates were incubated for an additional 4 h. MIC was defined as the lowest concentration at which no color change occurred (blue indicates no growth, while pink indicates microbial growth). For MBC/MFC determination, 20 µL from wells showing no visible growth was subcultured onto agar plates and incubated at 37 °C for 24 h (bacteria) or 25 °C for 48 h (C. albicans). The MBC/MFC was defined as the lowest concentration at which no colony growth was observed.

In silico analysis

Protein structure acquisition

The three-dimensional structures of penicillin-binding protein 2a (PBP2a) (Staphylococcus aureus), Kelch-like ECH-associated protein 1 (Keap1), and Melittin from Apis mellifera venom were retrieved from the RCSB Protein Data Bank (PDB) (https://www.rcsb.org/) using the following accession codes: 1MWT, 8X34, and 1BH1 for melittin, respectively (PDB, Protein Data Bank) (Figure 2). The PDB files were downloaded in “pdb” format for subsequent processing and docking analysis.

Proteins preparation

Prior to docking, both receptors (PBP2a or Keap1) and the ligand (melittin) proteins were prepared using AutoDockTools (ADT, version: 1.5.7) (Morris et al., 2009). Water molecules and heteroatoms not relevant to the interaction were removed from the PDB structures. Then, Hydrogen atoms were added to the protein structures to ensure correct ionization states at physiological pH. Moreover, Gasteiger charges were assigned to all atoms, and non-polar hydrogens were merged to optimize the structures for docking calculations.

Protein-protein docking

Protein-protein docking was performed using pyDockWeb tool (https://life.bsc.es/pid/pydockweb), a specialized docking and scoring algorithm designed for protein-protein interactions. pyDock employs a combination of electrostatic (Ele), desolvation (Desolv), and Van der Waals (VDW) energy terms to calculate the total binding energy according to the following formula:

The docking was carried out using default parameters (Jiménez-García et al., 2013). The top 10 poses were ranked by their total pyDock scores, with lower values indicating stronger binding affinity. For each receptor, the best-scoring pose was selected for subsequent interaction analysis.

To validate the pyDock predictions, independent docking was performed using ClusPro (Kozakov et al., 2017), a widely used protein-protein docking server (Alekseenko et al., 2020). The receptor and ligand PDB files prepared for pyDock were submitted to the ClusPro web server (https://cluspro.org) using default parameters. The top 10 clusters were ranked by cluster size, and the representative structure from the largest cluster was selected as the best-scoring model, which was then aligned to the pyDock-predicted pose in PyMOL to calculate RMSD and interface residue overlap.

Interaction visualization and analysis

Molecular visualization and detailed interaction analysis were performed using the PyMOL Molecular Graphics System (Version 3.1.0.4) (Schrödinger, LLC, New York, NY, USA). The best-scoring docking pose was loaded into PyMOL to analyze the protein-protein interface. Interacting residues were identified by selecting all atoms within a distance of 4.5 Å between the two proteins. Hydrogen bonds were defined with a cutoff distance of 3.5 Å, and salt bridges were identified between oppositely charged residues within 4.0 Å. Hydrophobic interactions were visualized by selecting non-polar residues at the interface (Laskowski and Thornton, 2022). Quantitative interface characterization was performed using PDBePISA (https://www.ebi.ac.uk/pdbe/pisa/pistart.html) (Krissinel and Henrick, 2007) to calculate buried surface area (BSA), solvation free energy (Δ i G), hydrogen bonds, and salt bridges at the protein-protein interface.

Analysis by size exclusion chromatography (FPLC)

Size-exclusion FPLC analysis of crude bee venom revealed a protein profile dominated by low-molecular-weight peptides, with only minor contributions from higher-molecular-weight proteins (Figure 3). The chromatogram displays a small, broad signal at low elution volumes corresponding to species >50 kDa, followed by weak, poorly resolved features in the intermediate region and a single dominant, sharp peak eluting in the peptide range. This overall pattern is consistent with the known HBV composition, which comprises a limited number of enzymatic proteins alongside a highly abundant pool of small bioactive peptides.

In the high-molecular-weight region (>50 kDa), the early shoulder exhibits low absorbance, indicating that large enzymes and glycoproteins constitute only a minor fraction of the total protein content, in agreement with proteomic studies of bee venom. Between the 50 kDa and 10 kDa calibration limits, only shallow, poorly resolved signals are observed, suggesting that proteins in this size range are present at relatively low abundance and cannot be confidently assigned to specific venom enzymes, such as phospholipase A 2 or hyaluronidase, based solely on size-exclusion behavior (Wehbe et al., 2019).

The chromatogram is dominated by a pronounced peak eluting in the low-molecular-weight region (<5 kDa), consistent with the expected elution range of melittin (≈2.8 kDa) and other short cationic peptides, including apamin (≈2.1 kDa) and MCD peptide (≈2.5 kDa), which together account for a substantial proportion of the venom dry mass. However, it should be noted that size-exclusion chromatography separates molecules based on hydrodynamic size rather than peptide identity; therefore, the composition of this dominant peak remains putative and cannot be unequivocally assigned to specific peptides without complementary identification techniques such as LC-MS/MS or MALDI-TOF MS. This FPLC profile is nevertheless consistent with a classic pattern of Apis mellifera venom, reflecting the predominance of low-molecular-weight bioactive peptides typical of this species (Wehbe et al., 2019). Proteomic studies consistently report melittin as a significant component of Apis mellifera venom, accounting for approximately 40%–50% of the total protein content (Gajski et al., 2024). The presence of a single dominant peak in the peptide region therefore suggests that the present extract may be enriched in melittin-like peptides, although direct confirmation by mass spectrometry would be required to establish peptide identity.

FTIR analysis

Fourier-transform infrared (FTIR) spectroscopy combined with attenuated total reflectance (ATR) was applied to elucidate the chemical features of dried AHBV. The FTIR spectra obtained from the three replicates collected showed a high degree of similarity, indicating the stability of the chemical fingerprint of the collected venom. The resulting FTIR-ATR spectrum (4000–400 cm -1) displayed a profile dominated by absorption bands characteristic of peptide and protein functional groups, confirming the protein-rich nature of the venom matrix (Figure 4). HBV is a complex biological secretion composed predominantly of peptides and enzymatic proteins, among which melittin is widely reported as the principal constituent (≈40–60% of dry weight), followed by phospholipase A 2 (≈10–12%), as reported in previous studies (Shi et al., 2022).

In the high-wavenumber region, a broad absorption extending from approximately 3600 to 3000 cm -1 was observed and is attributed to overlapping N–H and O–H stretching vibrations, reflecting extensive hydrogen bonding within peptide structures. (Kong and Yu, 2007). Additional absorptions at around 2960 and 2925 cm -1 correspond to asymmetric stretching vibrations of aliphatic–CH 2 groups, while weaker bands near 2870 and 2850 cm -1 are associated with symmetric C–H stretching modes (Barth, 2000; Tanuwidjaja et al., 2021).

The fingerprint region (1700–700 cm -1) exhibited distinct features typical of protein secondary structures. The dominant absorptions at approximately 1650 cm -1 (amide I) and 1540 cm -1 (amide II) arise from C=O stretching and N–H bending coupled with C–N stretching vibrations of peptide bonds, respectively, and are widely recognized as spectral markers of peptide-rich systems (Tanuwidjaja et al., 2021; Sandt, 2024). These bands are consistent with the reported prevalence of bioactive peptides and enzymes in HBV (Park et al., 2018).

Additional bands in the lower fingerprint region further support the proteinaceous composition of HBV. The absorption observed near 1080 cm -1 is attributed to C–H in-plane bending vibrations of aromatic amino acid residues in venom peptides. Weaker features detected around 805 cm -1 correspond to skeletal vibrations or C–N stretching modes within peptide backbones. Furthermore, absorptions near 730 and 660 cm -1 are assigned to amide V (N–H bending) and amide IV (O=C–N bending) vibrations, respectively, which are characteristic low-frequency indicators of protein secondary structure in peptide-rich biological materials (Tanuwidjaja et al., 2021).

Collectively, the FTIR-ATR analysis confirmed the proteinaceous and peptide-rich nature of AHBV, complementing the molecular weight distribution obtained by FPLC and the total protein content determined by the Bradford assay. Although FTIR does not enable the identification of individual peptide components, it provides complementary structural information that supports the overall biochemical characterization of the venom.

Protein content determination

Accurate determination of protein content is critical for bee venom proteome analysis, given the complex, peptide-rich nature of these secretions. In this study, crude AHBV was first dissolved at 1 mg/mL and then diluted in a series to prepare working solutions suitable for the Bradford assay. The total protein content of these solutions was estimated using the Bradford method and expressed as bovine serum albumin (BSA) equivalents, yielding a concentration of 204.8 ± 16.26 µg BSA equivalents/mL in the analyzed venom solution. This value reflects the protein concentration measured in solution after dilution and represents the dye-binding response relative to BSA. It does not correspond to the absolute protein mass fraction of dry crude venom, which is typically reported as a percentage (w/w) in the literature. Therefore, comparisons with studies expressing protein content as % (w/w) should be interpreted with caution.

The measured protein concentration of 204.8 ± 16.26 µg BSA equivalents/mL lies within the validated linear working range of the Bradford assay for venom samples (100–1500 μg/mL), indicating that the obtained value is analytically reliable and not affected by signal saturation or detection limits (Wanandy et al., 2024). The obtained protein concentration must be interpreted in the context of sample preparation and dilution. Literature reports on bee venom protein content vary considerably depending on whether values represent crude venom, concentrated samples, or diluted working solutions. For instance, Hoffman and Jacobson (1984) reported approximately 59 ± 7 µg protein per sting using Coomassie dye-binding methods, while Nazari et al. (2023) found that crude Apis mellifera venom contained approximately 56% protein (w/w) by Bradford assay. Consistent with previous comparative studies on HBV, Bradford-derived protein values calibrated with BSA provide a reliable estimate of total venom protein content. As with other colorimetric assays, these values are expressed as BSA equivalents, reflecting dye-binding behavior rather than absolute protein mass. Therefore, direct comparison of protein concentrations across studies employing different quantification methods or calibration standards should be interpreted with caution (Bocian et al., 2020; Wanandy et al., 2024). It should also be acknowledged that the Bradford assay may underestimate the protein content of melittin-rich samples, as the dye-binding response of this cationic peptide differs from that of bovine serum albumin (BSA) used as the calibration standard. Therefore, future studies could complement the present measurements using alternative protein quantification methods, such as the bicinchoninic acid (BCA) assay, to further validate total protein estimates (Wanandy et al., 2024).

Antioxidant activity

Bee venom exhibited pronounced radical-scavenging activity in the ABTS assay, reaching 727.0 ± 4.9 µmol Trolox equivalents (TE)/mg. The low standard deviation across replicates highlights excellent reproducibility and confirms the robustness of this measurement.

Few studies have quantified the antioxidant properties of whole bee venom using chemical radical-scavenging assays. Somwongin et al. (2018) characterized venoms from four Apis species (A. dorsata, A. mellifera, A. florea, and A. cerana). They reported measurable ABTS-based antioxidant activity, with A. dorsata venom exhibiting the highest capacity (10.21 mM TE/mg). Similarly, Denk (2023) observed ABTS scavenging by A. mellifera venom extracts across various solvents, confirming the radical-quenching potential of low-molecular-weight venom components (Somwongin et al., 2018; Denk, 2023).

Although direct quantitative comparisons across studies are limited by differences in species, assay methodology, venom extraction procedures, and test concentrations, the present results demonstrate strong and reproducible ABTS radical scavenging, consistent with previous reports indicating that crude bee venom possesses measurable antioxidant capacity (Martinello and Mutinelli, 2021). This activity aligns with the FPLC molecular weight profile, which is dominated by components below 3 kDa. This molecular weight range overlaps with that described in the literature for major bee venom peptides, such as melittin and apamin (Gajski et al., 2024). The predominance of this low-molecular-weight fraction suggests that small peptides may play a central role in the observed antioxidant activity, potentially through electron donation or hydrophobic interactions with ABTS + • radicals, mechanisms commonly described for peptide-rich extracts in Trolox-equivalent antioxidant assays (Carpena et al., 2020; Jung et al., 2022). The near absence of resolved 14–50 kDa protein peaks further indicates that higher-molecular-weight proteins are unlikely to be major contributors to the observed radical-scavenging activity (Danneels et al., 2015).

The pronounced radical-scavenging potential of the peptide-enriched fraction suggests that these low-molecular-weight components may also contribute to anti-inflammatory and protective effects against oxidative stress-related cellular damage, as previously reported for melittin and melittin-derived peptides (Scialò et al., 2017; Jung et al., 2022). This connection highlights the potential biomedical relevance of bee venom peptides as natural antioxidants and modulators of oxidative stress.

It should be noted that the ABTS radical scavenging activity was evaluated at a single concentration, which precluded reliable EC 50 determination. Future studies should establish a full dose-response curve across a wider concentration range to enable quantitative comparison with bee venom samples from different geographical origins. Furthermore, the present study evaluated antioxidant activity exclusively using the ABTS radical scavenging assay, which is based on a synthetic radical cation. While this assay provides a standardized and reproducible measure of antioxidant capacity, it does not fully reflect the complexity of biological oxidative stress. Future studies should therefore complement these findings by evaluating antioxidant activity against biologically relevant reactive oxygen species, including superoxide anion (O 2 • -), hydroxyl radical (•OH), and hydrogen peroxide (H 2 O 2), to provide a more comprehensive assessment of AHBV’s physiological antioxidant potential.

Antimicrobial activity of AHBV

Broth microdilution assays were performed to evaluate the antimicrobial activity of AHBV against selected Gram-positive and Gram-negative bacterial strains, as well as Candida albicans. MIC, MBC, and MFC values are summarized in Figures 5, 6. AHBV exhibited potent antibacterial activity against Gram-positive cocci. Staphylococcus aureus ATCC 6538P was the most susceptible strain (MIC 15.62 μg/mL; MBC 62.5 μg/mL), followed by S. aureus ATCC 25923 (MIC 62.5 μg/mL; MBC 125 μg/mL), while Bacillus subtilis ATCC 6633 was inhibited at 125 μg/mL (MBC 500 μg/mL). Generally, the MBC values for all tested pathogens were either equal to or higher than the MIC values.

In the present study, Gram-positive bacteria exhibited MIC values ranging from 15.62 to 125 μg/mL. Staphylococcus aureus ATCC 6538P showed the highest susceptibility, with an MIC of 15.62 μg/mL, which is lower than the value reported by Bakhiet et al. (2022) for the same strain (31.25 μg/mL). In addition, S. aureus ATCC 25923 displayed an MIC of 62.5 μg/mL, which remains lower than the MIC reported by Hegazi et al. (2014) for bee venom against S. aureus (1600 μg/mL), and comparable to the MIC of 12.5 μg/mL reported for S. aureus in studies on Anatolian HBV (Sonmez et al., 2022). Taken together, these values indicate that AHBV is highly effective against S. aureus, with activity close to that of Anatolian beevenom and markedly higher than Egyptian beevenom, underscoring its strong antibacterial potential.

Consistent with previous reports, Gram-positive bacteria were more susceptible to bee venom than Gram-negative species (El-Seedi et al., 2020; Sonmez et al., 2022). In the present study, Gram-negative bacteria exhibited higher MIC values, ranging from 125 to 500 μg/mL. Specifically, E. coli ATCC 8739 was inhibited at 125 μg/mL, whereas E. coli ATCC 25922 required 500 μg/mL, highlighting strain-dependent differences. For comparison, Hegazi et al. (2014) reported MIC values of 3600 μg/mL for E. coli, while Al-Ani et al. (2015) observed MICs of 200 μg/mL. AHBV also demonstrated moderate antifungal activity against Candida albicans ATCC 10231, with an MIC of 250 μg/mL and MFC of 500 μg/mL. These values align with previous reports, which showed MFCs between 62.5 and 125 μg/mL for clinical C. albicans strains (Lee, 2016), and 30–300 μg/mL for various Candida species, including C. parapsilosis, C. tropicalis, and C. krusei (Ratajczak et al., 2021). Differences among studies are likely due to variations in bee venom origin, collection method, and peptide content, and strain characteristics (Grinn-Gofroń et al., 2025). Geographical and ecological factors, such as climate and floral diversity, directly influence HBV composition and bioactivity, which may further contribute to the observed discrepancies between studies from different regions (Scaccabarozzi et al., 2021).

The semi-arid to Mediterranean climate of Sétif, with distinct seasons and local flora, may shape AHBV composition and bioactivity. Seasonal shifts in temperature, humidity, and floral resources can affect bee foraging and colony physiology, thereby influencing venom-gland activity and the levels of reportedly major peptides such as melittin and PLA 2 (Homrani et al., 2020; Scaccabarozzi et al., 2021). Consequently, these local factors may contribute to a venom profile relatively enriched in low-molecular-weight antimicrobial peptides, which is consistent with the relatively low MICs obtained for AHBV—particularly against S. aureus—compared with values reported for bee venoms from other regions (Hegazi et al., 2014; Sonmez et al., 2022).

FPLC analysis of our extract revealed a dominant low-molecular-weight fraction consistent with the peptide-rich composition of honeybee venom. However, because size-exclusion chromatography does not provide molecular identification, the identity of these peptides was not experimentally confirmed. Based on the literature, melittin and other low-molecular-weight cationic peptides are considered major contributors to the antimicrobial activity of honeybee venom and may therefore contribute to the biological activity observed in the present study (Leandro et al., 2015; Grinn-Gofroń et al., 2025). Among these peptides, melittin has been widely reported as a small, 26-amino-acid cationic peptide that is α-helical and amphipathic, enabling it to interact with and insert into microbial lipid bilayers. This interaction forms pores in the membrane, leading to leakage of cytosolic contents and ions, ultimately resulting in rapid cell lysis. The potency of this mechanism depends on the structure of the microbial cell envelope: Gram-positive bacteria, which lack an outer membrane, are more sensitive to melittin, whereas Gram-negative bacteria, though still susceptible, generally require higher concentrations for effective inhibition (Maitip et al., 2021). Peptides from bee venom have been shown to directly permeabilize fungal cell membranes, leading to leakage of essential cellular contents. For example, they exhibit fungicidal activity against Candida albicans by rapidly compromising membrane integrity and reducing cell viability (Lee, 2016). This moderate antifungal activity (MIC 250 μg/mL) may be associated with the membrane-disrupting properties of cationic peptides commonly found in honeybee venom. Whole bee venom is often reported to be more potent than purified peptides, suggesting possible synergistic effects among multiple venom components during microbial inhibition (Grinn-Gofroń et al., 2025).

Molecular docking of melittin

PBP2a–melittin molecular docking

The top 5 poses ranked by pyDock score are presented in Table 1. The best-scoring pose (Conf 3289) achieved a total score of (−40.406), with favorable electrostatic (−23.720) and desolvation (−17.127) contributions, and a low van der Waals term (+4.413) (Table 1), indicating proper steric complementarity without significant clashes.

Visualization of the best-scoring pose (Conf 3289) in PyMOL revealed that melittin binds within a structural pocket on chain A of PBP2a (Figure 7A). The binding site is distinct from the protein’s active site (Figure 7B).

To validate the pyDock predictions, independent docking was performed using ClusPro. The best-scoring ClusPro model was aligned to the pyDock-predicted pose. The two models showed a remarkably high degree of concordance, with an RMSD of 1.354 Å and an interface residue overlap of 94% (188 out of 200 aligned atoms). The interface residues identified by both methods were identical, confirming the robustness of the predicted binding mode. Analysis of the PBP2a-Melittin complex revealed an interaction interface involving 20 residues of PBP2a. The interface comprises negatively charged residues, polar residues, and hydrophobic residues (Table 2). Binding interface analysis revealed that melittin binds within a well-defined allosteric pocket on PBP2a chain A (Figure 8A). The binding pocket is formed by residues spanning positions 170–205 and 259–277. The pocket features a hydrophobic floor flanked by negatively charged residues (Table 2; Figure 8D).

The quantitative interface analysis using PDBePISA revealed a stable interface involving 40 residues of PBP2a and 22 residues of melittin (representing 84.6% of the melittin peptide), with a buried surface area of 901.3 Å 2 and a favorable solvation free energy gain of −10.9 kcal/mol (P-value = 0.265). The interface is stabilized by 3 hydrogen bonds and 2 salt bridges (Table 3; Figure 8D). These quantitative metrics are consistent with the pyDock scoring results, particularly the dominant electrostatic contribution (Ele = −23.720) and the extensive hydrophobic contacts.

Keap1-melittin molecular docking results

Molecular docking between Keap1 and melittin showed that the top-ranked pose (Conf 7366) achieved a total binding score of −31.537 kcal/mol, with favorable electrostatic (−27.640) and desolvation (−6.680) contributions. At the same time, the van der Waals term was positive (+27.828) (Table 4). These values indicate a thermodynamically favorable interaction dominated by electrostatic complementarity. Analysis of the best-scoring pose predicts that melittin binds to the exterior surface of the B chain of Keap 1, spanning residues 321–609 (Figure 9). In fact, this region corresponds to the Kelch domain of Keap1, which is the primary site for Nrf2 recognition (Ogura et al., 2010; Sinah et al., 2025).

Independent docking using ClusPro confirmed that melittin interacts with the same subunit of Keap1 and engages the same functional region as identified by pyDock. The RMSD between the best-scoring ClusPro model and the pyDock pose was 1.868 Å, with an interface residue overlap of 96.5% (193 out of 200 aligned atoms). While the ClusPro interface was more extensive, the convergence on the same functional site on the same subunit increases confidence in the predicted binding mode. The Keap1- Melettin interface comprises 12 residues from Keap1 and 11 residues from Melittin within a 4.5 Å cutoff (Table 1). Keap1 residues involved (Figures 10A,B), constitute the canonical Nrf2-binding pocket (Lo et al., 2006; Tong et al., 2006). Melittin interacts via its N-terminal region (residues 1–20), predominantly through hydrophobic contacts (Table 5) (Figure 10A).

Quantitative interface analysis revealed a modest interface for the Keap1-melittin complex. The PDBePISA analysis involved 16 residues of Keap1 and 12 residues of melittin (representing 46.2% of the melittin peptide) in interaction, with a buried surface area of 486.6 Å 2 and a modest solvation free energy gain of approximately −4.9 kcal/mol (P-value = 0.645). The interface is stabilized by 2 hydrogen bonds and one electrostatic interaction, with no true salt bridge detected (Table 3). The higher P-value (0.645) and the smaller buried surface area suggest that the predicted Keap1-melittin interaction, while structurally plausible, may be weaker than the PBP2a-melittin interaction.

The present docking study suggests that melittin binds to two functionally unrelated protein targets, PBP2a and Keap1, with different affinities, yet through distinct molecular mechanisms. Docking with PBP2a shows that melittin binds to an allosteric pocket on PBP2a with a score of −40.406 kcal/mol, which is higher than that observed for Keap1. The interaction is characterized by strong electrostatic complementarity, anchoring melittin to negatively charged residues surrounding the active site. This binding site corresponds to a previously characterized allosteric regulatory region located approximately 60 Å from the catalytic domain (Otero et al., 2013; Tabassum et al., 2023). The extensive interface (20 PBP2a residues, buried surface area ∼950 Å 2) and favorable desolvation energy suggest a stable complex. These findings support a model whereby is predicted to bind to a region consistent with the allosteric pocket of PBP2a. By binding to this site, melittin may induce conformational changes that open or destabilize the active site, potentially resensitizing methicillin-resistant Staphylococcus aureus (MRSA) to β-lactam antibiotics (Mahasenan et al., 2017). This provides a molecular rationale for melittin’s known antimicrobial activity against MRSA (Raghuraman and Chattopadhyay, 2007; Choi et al., 2015; Oduniton et al., 2024).

Additionally, melittin binds to the Keap1 Kelch domain, yielding a total score of −31.537 kcal/mol, engaging residues critical for Nrf2 recognition. Notably, Trp403 and Trp450, which form a hydrophobic pocket essential for Keap1-Nrf2 binding (Lo et al., 2006; Fukutomi et al., 2014; Sinha et al., 2025) are directly involved in melittin binding. The interaction is mediated primarily by hydrophobic contacts involving the N-terminal region of melittin, with no contribution from salt bridges. This binding mode resembles that of the Nrf2 ETGE motif, suggesting a mechanism of competitive inhibition. Given that Keap1 functions as a negative regulator of the Nrf2-mediated antioxidant response (Hayes and Dinkova-Kostova, 2014; Baird and Yamamoto, 2020), melittin binding to the Kelch domain is predicted to disrupt Keap1-Nrf2 interaction, leading to Nrf2 stabilization, nuclear translocation, and subsequent activation of cytoprotective gene expression.

Although melittin binds both targets with favorable energetics, the PBP2a interaction is stronger (−40.406 vs. −31.537 kcal/mol) and dominated by electrostatic forces. The weaker thermodynamic profile of the Keap1-melittin interface (buried surface area 486.6 Å 2, Δ i G ≈ −4.9 kcal/mol) compared to the PBP2a-melittin complex (901.3 Å 2, Δ i G = −10.9 kcal/mol) can be rationalized by the distinct binding modes of melittin on these two targets. In the case of PBP2a, melittin is deeply buried within a well-defined allosteric pocket, establishing extensive three-dimensional contacts with 40 residues of the protein. In contrast, the predicted Keap1-melittin interaction involves the adhesion of melittin onto the exterior surface of the Kelch domain, engaging only 16 residues of Keap1.

Importantly, melittin engages these targets through distinct regions of its structure, involving its positively charged C-terminal segment for PBP2a and its hydrophobic N-terminal helix for Keap1, thereby demonstrating structural plasticity. However, it should be noted that melittin undergoes C-terminal amidation, a post-translational modification that increases its net positive charge and enhances its amphipathic character and membrane activity (Raghuraman and Chattopadhyay, 2007; Chaudhary et al., 2025). Although the present docking analysis employed the unmodified peptide, the C-terminal amidation may contribute to the electrostatic interactions observed in the PBP2a complex, potentially stabilizing the binding pose. Future studies incorporating the amidated form of melittin in molecular dynamics simulations could further clarify this contribution.

Finally, this molecular docking provides structural hypotheses and plausible binding models, but docking alone cannot establish binding affinity, allosteric modulation, competitive inhibition, or downstream biological activity under physiological conditions. The predicted binding modes should therefore be regarded as hypothesis-generating rather than conclusive evidence of molecular mechanism and require validation through complementary biophysical and functional studies. Furthermore, the present molecular docking analysis focused exclusively on melittin because it is widely recognized as one of the principal bioactive peptides of honeybee venom. However, the in vitro antimicrobial and antioxidant activities were evaluated using whole honeybee venom, whose biological effects are likely to result from the combined action of multiple peptides, proteins, and other bioactive constituents. Therefore, the proposed molecular mechanisms should be interpreted as representing one potential contributor to the observed biological activities rather than a complete explanation of the biological effects of the whole venom. Possible synergistic or antagonistic interactions among venom components were beyond the scope of the present study and deserve further investigation. In addition, although melittin represents a promising bioactive peptide, its well-documented cytotoxic and hemolytic properties should be carefully considered before therapeutic or food-related applications can be envisaged. Future studies should therefore evaluate both efficacy and safety in appropriate biological models.

This study demonstrated that Algerian honeybee venom contains a predominant low-molecular-weight peptide fraction, together with substantial protein content, and exhibits antioxidant and antimicrobial activities. The characterization of honeybee venom from Northeastern Algeria—a region with limited compositional and functional data—expands current knowledge of honeybee venom diversity and provides region-specific data valuable for future comparative and applied research. Complementary in silico molecular docking was performed using melittin, a major peptide widely reported in Apis mellifera venom, to explore its potential interactions with PBP2a and Keap1. The docking analyses predicted energetically favorable binding poses for both targets. On PBP2a, melittin occupies an allosteric pocket via salt bridges and hydrophobic contacts, suggesting potential as an allosteric inhibitor that may resensitize MRSA to β-lactam antibiotics. On Keap1, melittin engages the Kelch domain via hydrophobic interactions involving critical Nrf2-binding residues, hypothesizing a competitive inhibition mechanism that could activate the Nrf2 antioxidant pathway. Together, these computational findings generate mechanistic hypotheses regarding the multifunctional bioactivity of melittin and expand the understanding of honeybee venom as a natural source of bioactive peptides. Overall, these results suggest that melittin represents a promising bioactive candidate for food safety and preservation applications, particularly in controlling microbial contamination and oxidative processes. Further experimental validation—including surface plasmon resonance, isothermal titration calorimetry, and cellular assays—is warranted to confirm these interactions and explore the potential applications of melittin and its derivatives as functional bioactive agents.