Introduction

Infectious diseases have become a pressing global concern due to the emergence and spread of drug-resistant bacterial strains, leading to increased treatment costs and the urgent need for new and more effective antibiotics1,2. The current problem is particularly acute in developing countries such as Vietnam, where the rise of antimicrobial resistance significantly hampers the management of bacterial infections. Staphylococcus aureus, one of the most prevalent infectious pathogens worldwide, causes a wide spectrum of infections ranging from mild skin conditions to life-threatening systemic diseases3,4,5. Interestingly, a study involving 293 S. aureus isolates revealed that 49.7% were methicillin-resistant Staphylococcus aureus (MRSA), underscoring the urgent need for novel therapeutic strategies6,7. To address this challenge, research has increasingly focused on antibacterial compounds derived from plants, which often offer broad-spectrum activity, tend to have fewer side effects, and lower resistance potential compared to synthetic antibiotics8,9.

Sortase A (SrtA) is a key enzyme in the transpeptidase family, primarily found in gram-positive bacteria. This enzyme plays a pivotal role in anchoring surface proteins to the bacterial cell wall, a process essential for the survival and virulence of many pathogens. First discovered in the late 1990 s in Staphylococcus aureus, SrtA quickly became a subject of interest in microbiology and pharmaceutical researches. In S. aureus, SrtA is a protein consisting of 206 amino acids that includes a signal peptide and a membrane anchor at the N-terminus10. SrtA functions as a housekeeping protein that plays a crucial role for bacterial physiology11,12. This enzyme catalyzes the covalent attachment of proteins containing the LPXTG motif to the bacterial cell wall10. The bacterial adhesion process begins with SrtA recognizing the LPXTG motif on microbial surface components known as MSCRAMMs (microbial surface components recognizing adhesive matrix molecules). Following this recognition, SrtA then performs two main reactions: thioester formation and peptide transfer13. In the first step, the catalytic cysteine residue (Cys184) of SrtA cleaves the LPXTG motif between the threonine and glycine residues, forming an acyl-enzyme thioester intermediate. This activation is facilitated by the imidazole ring of His120, which enables the formation of a thiolate13. The thioester intermediate is subsequently resolved by nucleophilic attack from the N-terminus of the penta-glycine (Gly5) chain in the bacterial peptidoglycan. Arg197 supports this cleavage reaction by stabilizing the tetrahedral oxyanion transition state and providing the necessary energy for catalysis14. The peptide transfer reaction results in the formation of an amide bond between the C-terminal threonine of MSCRAMM and the Gly5 chain of peptidoglycan, effectively anchoring the protein to the bacterial cell wall. Inhibition of any step in this process disrupts functions of SrtA, preventing the anchoring of the MSCRAMM to peptidoglycan, therby reducing the ability of bacteria to adhere to host tissues, invade cells, and evade immune response. SrtA is an important protein in the toxicity of many pathogenic bacteria in their hosts. Because of its role in virulence rather than viability, SrtA has emerged as a promising therapeutic target, especially for infections caused by multidrug-resistant bacteria such as MRSA (methicillin-resistant Staphylococcus aureus). Owing to the strong antibiotic resistance of MRSA, the development of alternative treatment methods, such as SrtA inhibition, may provide long-term effectiveness in controlling these infections15. SrtA is considered a potential target with therapeutic roles in many diseases caused by bacterial species. The inhibition of SrtA may help prevent the development of drug resistance, as the mechanism of action involves pathogenicity rather than bacterial survival. Given the rising threat of global antibiotic resistance, targeting SrtA may offer an effective alternative strategy. Unlike conventional antibiotics, SrtA inhibitors may exert less selective pressure for resistance development, making them attractive candidates for long-term infectious disease management14,15.

In our screening for antibacterial agents from Vietnamese natural and medicinal plants, we found that Jatropha podagrica Hook collected in Vietnam showed as a promising candidate with notable activity against S. aureus and MRSA. Although J. podagrica is a native to tropical Central America, it is now widely cultivated in other tropical and subtropical regions, including Africa, North America, and parts of Asia16. Originally from the Caribbean, this plant was introduced to other regions as an ornamental species due to its year-round red blossoms and adaptability. In Vietnam, it is commonly grown as a decorative plant throughout the country17. Since the 1980 s, scientists have explored the chemical composition and biological activities of J. podagrica. A variety of phytochemicals, including japodic acid, fraxidin, fraxetin, scoparone, 3-acetylaleuritolic acid, β-sitosterol, sitosterol, flavonoids, and diterpenoids, have been isolated from plant stem bark16,18,19,20. However, the biological activities of these compounds remain largely unexplored21,22. While several bioactive compounds have been isolated from different parts of Jatropha podagrica, studies of its antibacterial properties, particularly against S. aureus and MRSA strains in Vietnam, are still limited. Therefore, this study aims to evaluate the antibacterial properties of extracts and isolated compounds present in Jatropha podagrica Hook against S. aureus and MRSA through in vitro analysis, and predict the antibacterial mechanism of these compounds by targeting sortase A (SrtA) via in silico methods.

Results

In vitro antibacterial activity

The fresh leaves, stems, and roots of J. podagrica were dried in the shade and then ground into powder. Detailed information regarding the extract samples and fractions is presented in Table S1. The antibacterial activity of the root, stem, and leaf extracts of J. podagrica was assessed via the agar diffusion method. The effects of the crude ethanolic extract, n-hexane, and ethyl acetate fractions were tested against S. aureus and MRSA pathogens. The inhibitory effect on bacterial growth was determined by measuring the diameter of the inhibition zone (ZOI) (Table 1).

Table 1 Zone of Inhibition (mm) of J. podagrica extracts on S. aureus and MRSA strains.
Full size table

As shown in Fig. 1, both S. aureus and MRSA were sensitive to the root and stem extracts, as indicated by the formation of inhibition zones on the agar plates. In contrast, the leaf extract showed no antibacterial activity. The n-hexane fraction of the root extract (100 mg/mL) had the greatest inhibitory effect, with inhibition zone diameters of 11.33 ± 0.33 mm for S. aureus and 12.00 ± 1.00 mm for MRSA. The ethyl acetate fraction at the same concentration had a lower antibacterial effect, with inhibition zone diameters of 8.33 ± 0.21 mm and 11.00 ± 1.00 mm. No inhibition zones were observed in the negative control wells containing 10% DMSO, confirming that the solvent did not interfere with the antibacterial activity of the extracts. These findings initially confirmed that the root extracts of J. podagrica can suppress the growth of S. aureus and MRSA strains isolated from clinical samples. Therefore, the root extracts were chosen for further experiments to investigate their antibacterial activity (Fig. S1).

MIC and MBC results

The MIC values of the crude extracts and fractions obtained from J. podagrica roots were identified based on the color change from blue to pink of the resazurin (0.015%). A pink color was observed in the wells supplemented with equal to or less than 3.125 mg/mL ethanolic extracts, suggesting that S. aureus and MRSA were not inhibited at these concentrations. Moreover, the MIC of the ethanolic extract was 6.25 mg/mL (Fig. 2A). The MICs of the ethyl acetate and n-hexane fractions were 0.78 mg/mL (Fig. 2B) and 0.20 mg/mL (Fig. 2C), respectively. Furthermore, the color in the wells containing only MHB and different amounts of extract or fractions remained blue, indicating that neither MHB nor extract/fractions interfered with the color change of the indicator solution.

The minimum bactericidal concentrations (MBCs) of these extracts and fractions were then determined by evaluating the growth of the targeted bacteria after overnight incubation with the extract at concentrations of 1 × MIC, 2 × MIC, 4 × MIC, and 8 × MIC. The MIC and MBC values for the J. podagrica root extracts revealed that the ethanolic extract and ethyl acetate fraction have bactericidal efficacy against S. aureus and MRSA. The n-hexane fraction was considered a bacteriostatic agent for these two pathogens (Table 2 and Fig. S1).

Table 2 MIC and MBC results of the roots of J. podagrica.
Full size table

Molecular docking analysis

The in vitro results indicated that extracted fractions containing potential compounds, particularly from the roots, could inhibit S. aureus and MRSA. To explore this, we employed bioinformatics techniques to predict the mechanism of action of 43 compounds23 against the target protein sortase A (PDB ID: 1T2P). Table 3 lists the top 10 compounds with the highest binding affinities among the 43 compounds from J. podagrica. For sortase A (1T2P), luteolin showed the strongest binding affinity at −9.0 kcal/mol, followed by isovitexin (−8.7 kcal/mol), 7,4’-dimethoxyflavone (−8.6 kcal/mol), and acacetin (−8.6 kcal/mol). Luteolin forms hydrogen bonds with the amino acids Gln172, Gln178, and Lys175, which are essential for stabilizing the ligand in the protein’s active site. Additionally, luteolin forms hydrophobic (van der Waals) interactions with the amino acids Ile158, Val166, Pro163, and Leu169, enhancing its binding. Pi‒Sigma interactions with Thr180 and Ile199, along with Pi‒alkyl interactions with Val168 and Val201, further suggest nonspecific stabilization (Fig. 3). These combined interactions imply that luteolin can stably bind to SrtA, potentially inhibiting its enzymatic activity.

Table 3 Top 10 binding energy compounds for 1T2P.
Full size table

Molecular dynamics of the binding of luteolin to sortase A (1T2P)

The ability of luteolin to bind to the sortase A protein was further assessed via dynamic modeling. The RMSD plot of the protein’s Cα atoms (Fig. 4A) showed a wide range of values from 0 to 100 ns, varying between approximately 1.5 Å and over 3.0 Å. The RMSD graph indicates that while the protein underwent fluctuations, it reached relative stability after approximately 20 ns. Between 40 and 80 ns, the fluctuations were more obvious, suggesting potential structural adjustments during this period. The luteolin RMSD was lower, ranging from 0.6 Å to 1.6 Å, with slight increases and decreases (Fig. 4A). The relatively low and stable RMSD values after the initial phase suggest that the ligand remained stable at the binding site. These observations indicate that while the protein structure experienced some degree of fluctuation and adjustment, the ligand was able to maintain a stable position within the binding pocket over time. This stability supports the potential of luteolin to interact effectively and consistently with sortase A.

The RMSF plot (Fig. 4B), which shows the fluctuation of individual amino acids in the protein chain, provides additional insights. High RMSF values indicate flexible or fluctuating regions, whereas low RMSF values denote more stable regions. In this plot, significant peaks were observed, particularly between residues 100 and 120, with the highest peak exceeding 4.0 Å. This finding suggests that these regions are highly flexible, potentially corresponding to loops or segments not part of a stable secondary structure. More stable regions, with RMSF values less than 1.5 Å, likely correspond to α-helices or β-sheets, which are more rigid components of the secondary structure. Regions with high RMSF values might indicate flexible sites that are crucial for protein function, such as docking areas or interaction sites for ligands or other partners. Conversely, stable regions could represent the protein’s core structures that help maintain overall integrity during luteolin binding. The high fluctuations in specific regions may reflect necessary structural adjustments to accommodate luteolin binding.

Protein and ligand contact histogram and timeline

To further assess the dynamic interaction behavior of luteolin with SrtA, we performed a protein–ligand contact analysis over the course of a 100 ns molecular dynamics simulation (Fig. 5). The interaction profile revealed that luteolin consistently formed a broad spectrum of stabilizing contacts with the SrtA binding pocket, including hydrogen bonds, water bridges, hydrophobic interactions, and ionic contacts (Fig. 5A). Among these, hydrogen bonds and water bridges were the predominant interaction types, suggesting that polar and solvent-mediated contacts play a central role in maintaining the stability of the luteolin–SrtA complex. Key residues such as Glu171, Gln172, Asn114, Gln178, and Lys173 demonstrated high interaction frequencies, indicating that these amino acids form persistent contacts with luteolin throughout the simulation. These interactions contribute significantly to the spatial orientation and retention of luteolin within the enzyme’s active site. Additionally, hydrophobic interactions, although less frequent, were observed with residues lining the hydrophobic core of the binding pocket, likely stabilizing the aromatic rings of the luteolin scaffold. Furthermore, a small number of ionic interactions involving positively charged residues (e.g., Arg197) provided electrostatic stabilization, complementing the polar interaction network.

Figure 5B presents the timeline of interactions between luteolin and sortase A over a 100 ns simulation period. The total number of interactions varies between approximately 3 and 9, indicating fluctuations and transient plateaus throughout the timeframe. These variations may correspond to changes in the position or stability of luteolin within the protein binding pocket. The lower portion of the timeline shows specific residues that interact with luteolin over time, with colors ranging from light to dark, indicating an increasing number of interactions (from 1 to ≥ 4). Certain residues maintain continuous interactions throughout the simulation, suggesting their importance in stabilizing luteolin within the protein. In contrast, disruptions or reductions in interactions may point to shifts in ligand positioning or structural changes in the protein. Biologically, residues with continuous and strong interactions (shown as darker colors with many points along the time axis) are critical anchors for luteolin within the binding site. The periods in which the total number of interactions decreases could indicate either ligand movement or conformational changes in the protein structure.

Ligand and protein properties in dynamic action

Figure 6A shows the ligand root mean square fluctuation (L-RMSF) for each atom in the luteolin structure when bound to sortase A. The average RMSF for the atoms was relatively high, approximately 25 Å, suggesting significant variability. This implies that certain parts of luteolin strongly fluctuate or do not maintain a stable position in the binding pocket throughout the simulation. The fairly uniform variability across all atoms suggests that the entire ligand structure may exhibit flexibility or displacement when interacting with the protein. This substantial fluctuation indicates that luteolin might not be tightly bound or is unstable under the current simulation conditions, potentially impacting its interaction efficiency with sortase A. Further analyses or adjustments to simulation conditions may be needed to increase its stability. Figure 6B highlights the ligand properties throughout the simulation. The RMSD of luteolin varied between 0.5 Å and 1.5 Å, indicating that the ligand maintained a degree of stability within the protein binding pocket. Occasional periods of larger fluctuations might reflect adjustments in the position or structure of luteolin. The radius of gyration, ranging from 3.45 Å to 3.75 Å, suggested minimal changes in the folding or expansion of the ligand structure, indicating that luteolin preserved its overall conformation without significant expansion or contraction. Intra-HB analysis revealed that the number of intramolecular hydrogen bonds fluctuated between 0 and 1, implying that luteolin occasionally maintains one or no internal hydrogen bonds, impacting its rigidity or flexibility. The Molecular Surface Area (MolSA) ranged from 265 to 275 Ų, indicating the stability of the surface area of luteolin relative to its environment. The surface area exposed to solvent (SASA) fluctuated from approximately 150 Ų to over 450 Ų, with significant decreases indicating deeper embedding of luteolin into the binding pocket or reduced exposure to the solvent. The polar surface area (PSA) ranged from 210 to 224 Ų, indicating stable polar exposure throughout the simulation, which may support hydrogen bonding or electrostatic interactions with the surrounding environment.

Despite their general stability, properties such as RMSD, intra-HB, and SASA fluctuated, suggesting positional adjustments and dynamic interactions between luteolin and sortase A. Figure 6C shows the % SSE (secondary structure element) plot for each protein residue, indicating the percentage of time each residue retained its secondary structure (e.g., alpha-helix, beta-sheet) throughout the simulation. The light blue regions represent residues that retained stable secondary structures nearly 100% of the time, indicating high stability. Conversely, red or near-red regions indicate residues that frequently changed their secondary structures or did not maintain a stable structure, indicating flexibility or instability. The most stable regions, such as residues 10–20, 60–80, and 120–150, represent the structural core of the protein, which is essential for maintaining its integrity and function. Certain segments, such as those near residues 40 and 100, displayed low secondary structure retention, suggesting high flexibility or the presence of unstable loops. The flexible region near residue 40, likely a loop or nonstructural segment, could facilitate spatial adjustments for ligand interactions or support necessary protein movements. If luteolin interacts with these regions, the flexibility could allow adjustments for efficient binding. The region around residue 100 also exhibited frequent secondary structure changes, indicating flexibility that might be crucial for functional adjustments during ligand interactions, as evidenced by decreased retention in the SSE timeline (Fig. 6D).

Luteolin torsion analysis

Luteolin, a flavonoid, has hydroxyl (OH) groups positioned on various parts of its rings, which play significant roles in hydrogen bonding—both intramolecularly and intermolecularly (with proteins). The benzopyran rings and their connecting bridges contribute to the characteristic structure, stability, and binding ability of luteolin. Torsion angles in the luteolin structure influence its flexibility and conformational adaptability during interactions with proteins. These torsion angles determine how the rings and functional groups can be spatially repositioned to enhance interactions with proteins (Fig. 7). Variations in these angles may affect the potential of luteolin to form hydrogen bonds or hydrophobic interactions with residues within the sortase A binding pocket. The flexible hydroxyl groups and torsion angles enable luteolin to adjust its position for optimal protein interactions, aligning with the stability and adaptability observed in molecular dynamics simulations.

Figure 7 shows the histogram of torsion angle distributions for the bonds within luteolin during the simulation. Each row represents a specific torsion angle and its distribution across values from − 180° to 180°. A broad distribution indicates high flexibility, whereas a concentration at specific angles suggests stability at those torsion angles. The extent of torsion angle flexibility can indicate how luteolin adjusts its conformation within the sortase A binding pocket. Widely distributed torsion angles suggest that luteolin can adopt different conformations to optimize protein interactions, whereas a concentrated distribution implies that some bonds maintain a stable structure, enhancing the stability of the ligand-protein complex. The Ramachandran-like plots depict the frequency of torsion angle values in 2D space, highlighting where torsion angles commonly occur within the molecule. Scattered points represent highly flexible torsion angles, whereas clusters indicate stable angle values throughout the simulation. These clusters suggest that certain parts of luteolin maintain a stable conformation, whereas areas with a broader distribution indicate the ability of the molecule to adjust its structure according to the binding environment. This flexibility allows luteolin to tailor its shape for improved interactions with sortase A, facilitating stable ligand-protein binding.

Determination of the MIC and MBC of luteolin against SA and MRSA

Our in silico study revealed that luteolin had a high binding affinity for sortase A. This finding suggested that luteolin exerts antibacterial activity against S. aureus and MRSA by interacting with sortase A. To further support the antibacterial results and docking simulations, we isolated luteolin from the roots of J. podagrica and carried out in vitro antibacterial activity of luteolin, which was subsequently validated based on MIC and MBC values (Fig. 8).

At a dose of 62.5 µg/mL, luteolin effectively inhibited the growth of S. aureus and MRSA within 24 h (Fig. 8A), suggesting that the MIC value of luteolin against S. aureus and MRSA was at 31.25 µg/mL. Furthermore, MBC was determined based on the co-culture of S. aureus and MRSA in the presence of 1xMIC, 2xMIC, and 4xMIC, respectively. The MBC/MIC ratio was greater than 4.0, demonstrating a bacteriostatic effect on the target pathogens (Fig. 8B). Xi et al. (2022) reported that luteolin at concentrations equal to or greater than 312.5 µg/ml reduced S. aureus proliferation24. According to Sun et al. (2022), the MICs of luteolin against S. aureus ATCC 29,213, ATCC25923, MRSA N315, and S. aureus isolated from inpatients range from 64 µg/mL to 128 µg/mL25. In our study, luteolin had a stronger effect on S. aureus, with a lower MIC value (62.5 µg/ml). Luteolin has been reported to inhibit biofilm formation and promote morphological alterations in S. aureus and MRSA26. Luteolin also decreased the cytotoxicity of MRSA by reducing the levels of a-hemolysin, d-hemolysin, and hlaA in MRSA25.

Absorption, distribution, metabolism, excretion, and toxicity (ADMET) predictions analysis

In addition to the antibacterial and mechanistic evaluations, we conducted an in silico ADMET (absorption, distribution, metabolism, excretion, and toxicity) analysis of luteolin to preliminarily assess its pharmacokinetic behavior and drug-likeness. The physicochemical profiling revealed favorable characteristics, including a molecular weight of 286.24 g/mol, moderate lipophilicity (log P = 1.73), and a total polar surface area (TPSA) of 111.13 Ų—parameters which fall within acceptable ranges for good oral bioavailability. Luteolin also demonstrated high predicted gastrointestinal absorption and did not violate Lipinski’s rule of five, supporting its potential suitability for oral administration. Despite these promising parameters, the ADME analysis highlighted several metabolic considerations. Luteolin is predicted to act as an inhibitor of cytochrome P450 isoenzymes CYP3A4 and CYP2D6, which play major roles in hepatic drug metabolism. This suggests a potential risk of drug–drug interactions when co-administered with substrates of these enzymes and warrants further pharmacokinetic and enzymatic validation. Moreover, the compound is not predicted to cross the blood–brain barrier, which may limit its use in central nervous system-related infections but could be advantageous in reducing off-target neurotoxicity. Regarding toxicity, the DL-AOT model predicted an acute oral LD₅₀ of 3.17 mg/kg for luteolin, classifying it within the cautionary range. While this value does not preclude its therapeutic application, it does emphasize the importance of dose titration and the need for thorough in vivo toxicological assessment in future studies. Furthermore, although luteolin exhibited strong bioactivity and binding affinity in vitro and in silico, its moderate predicted skin permeability (log Kp = − 6.25 cm/s) may limit its effectiveness in transdermal formulations (Tables S3-S5). Taken together, these ADMET predictions suggest that luteolin possesses drug-like properties suitable for further development as an orally active antibacterial agent. However, its potential metabolic liabilities and predicted toxicity profile highlight the necessity for comprehensive pharmacological, toxicokinetic, and formulation studies before clinical translation.

Discussion

This study explored the antibacterial potential of Jatropha podagrica extracts, with a particular emphasis on luteolin—a flavonoid identified as a key bioactive constituent—against Staphylococcus aureus and methicillin-resistant S. aureus (MRSA). The investigation integrated both in vitro assays and in silico modeling to elucidate the compound’s mechanism of action, particularly its interaction with the sortase A (SrtA) enzyme, a validated virulence factor in Gram-positive bacteria.

Our findings are consistent with previous reports on luteolin’s broad-spectrum antimicrobial properties. Xi et al. (2022) demonstrated that luteolin inhibited S. aureus with MIC values similar to our results (31.25 µg/mL), and additionally showed membrane-disruptive effects and inhibition of biofilm formation24. Ding et al. (2024) further highlighted luteolin’s potent anti-biofilm activity against multidrug-resistant (MDR) E. coli, achieving over 87% inhibition at 2× MIC. These outcomes align well with our in vitro findings, confirming luteolin’s efficacy not only against S. aureus but also against MRSA, a clinically significant pathogen with limited therapeutic options27.

Mechanistically, our docking simulations revealed a strong binding affinity between luteolin and the active site of SrtA, specifically involving the catalytic residues His120, Cys184, and Arg197—consistent with the enzymatic interaction model described by Suree et al. (2009)28. The compound formed multiple stabilizing hydrogen bonds with residues such as Gln172, Lys173, and Gln178 within the substrate recognition loop, along with hydrophobic contacts involving Val166, Ile158, and Ile199. Additional π-alkyl and π–σ interactions further stabilized the flavonoid core within the hydrophobic cleft. These interactions support a structurally favorable and functionally relevant binding pose, corroborated by our 100-ns molecular dynamics (MD) simulation, which demonstrated sustained protein–ligand stability with minimal RMSD fluctuation.

Compared to previous studies where luteolin’s antimicrobial role was assessed in commercial or isolated systems, our work is the first to report the antibacterial activity of J. podagrica extracts from Vietnam. Notably, the root extract exhibited the strongest activity (MBC/MIC = 2.0), prompting further characterization and confirmation of luteolin as a major active compound in the flowers, stems, and roots. In vitro antibacterial assays with the isolated compound reaffirmed its inhibitory effect on both S. aureus and MRSA. Despite exhibiting a bacteriostatic mode of action (MBC/MIC > 4.0), luteolin’s ability to suppress bacterial growth and biofilm formation renders it therapeutically valuable, especially in mitigating resistance development often associated with bactericidal agents.

Beyond the mechanistic insights, our study adds to the growing body of evidence supporting the potential application of plant-derived flavonoids in infectious disease management. Silkar et al. (2022) previously reported that incorporating luteolin into hand sanitizers significantly enhanced antimicrobial efficacy, yielding inhibition zones against S. aureus up to 32.9 mm—outperforming standard commercial formulations24. These translational findings, together with our molecular-level validation, underscore luteolin’s applicability in both preventive and therapeutic settings.

To complement the biological and mechanistic assessments, we conducted an in silico ADMET analysis to evaluate luteolin’s drug-likeness and pharmacokinetic properties. The physicochemical profiling revealed favorable characteristics, including a molecular weight of 286.24 g/mol, moderate lipophilicity (log P = 1.73), and high gastrointestinal absorption, with no violations of Lipinski’s rule of five, suggesting good oral bioavailability. However, ADME predictions indicated potential metabolic liabilities, as luteolin is predicted to inhibit CYP3A4 and CYP2D6, raising the risk of drug–drug interactions. Additionally, its inability to cross the blood–brain barrier may limit use in CNS infections, while reducing neurotoxicity risk. The predicted LD₅₀ value of 3.17 mg/kg classifies luteolin as a compound requiring caution in dose escalation. Although the predicted skin permeability was low (log Kp = − 6.25 cm/s), the overall ADMET profile supports further development of luteolin as a candidate for oral antibacterial therapy, provided its metabolic and toxicological parameters are validated experimentally.

Despite these promising outcomes, certain limitations should be acknowledged. First, our in silico analysis did not include an apo (protein-only) MD simulation for SrtA, as previous studies have already characterized the unbound conformational dynamics of this enzyme under similar conditions28,29. Our focus was to analyze the complex form to capture ligand-specific dynamics. Second, in vitro assays were limited to planktonic cultures, without evaluating biofilm models or co-culture conditions, which may affect clinical translatability. Third, the absence of in vivo data restricts our ability to predict pharmacokinetics, bioavailability, and systemic safety. Lastly, the natural abundance, extraction yield, and formulation stability of luteolin were not optimized, which could influence its development into therapeutic products. However, by situating our findings within the context of existing literature, this study provides a comprehensive understanding of luteolin’s antibacterial potential, both from ethnobotanical and mechanistic perspectives. The integration of experimental and computational methods highlights luteolin as a promising lead for future anti-MRSA drug development. Ongoing and future work will focus on expanding bioactivity testing under more complex biological models, formulation enhancement, and preclinical in vivo validation to advance its translational potential.

Conclusions

In conclusion, based on the in vitro experimental results, the extracts from the stem and root of the Vietnamese Jatropha podagrica Hook exhibited antibacterial activity against S. aureus and MRSA, whereas the leaf extracts did not inhibit either strain. The n-hexane fraction from the roots presented a MIC value of 0.2 mg/mL and an MBC value of 3.13 mg/mL. The MBC/MIC ratio confirmed the antibacterial activity, with the total ethanol extract and ethyl acetate fraction displaying bactericidal properties, whereas the n-hexane fraction exhibited bacteriostatic properties. In silico analysis revealed that luteolin strongly binds to an antibacterial target protein, sortase A. Molecular dynamics analysis indicated that luteolin maintained stable and robust interactions with sortase A, which was supported by structural stability, torsional angle flexibility, and strong ligand-protein interactions. Additionally, the in vitro test results demonstrated that luteolin is a potential candidate for further investigations in anti-Staphylococcus aureus activity. These findings provide the first scientific report on the antibacterial activity of Vietnamese J. podagrica, suggesting a potential direction for developing new drugs capable of aiding in the treatment of S. aureus and MRSA infections.

Materials and methods

Microorganisms

Staphylococcus aureus (S. aureus) and methicillin-resistant Staphylococcus aureus (MRSA) strains were isolated from the pus of a skin-infected patient in previous studies conducted by our research group30. The bacteria were activated, enriched, and preserved in Mueller‒Hinton (MH) medium at the Laboratory of the Department of Biotechnology, Faculty of Chemical Engineering, University of Science and Technology, the University of Danang, Vietnam.

Plant collection and extraction process

Jatropha podagrica Hook. was collected in October 2022 from the Phu Lap commune, Tan Phu district, Dong Nai Province. The plant voucher specimen was identified by Loi Huynh PhD, and the samples were stored in the Laboratory of the Department of Pharmacognosy and Drug Control, School of Medicine and Pharmacy, the University of Danang, Vietnam. The fresh leaves, stems, and roots of J. podagrica were shade-dried and ground into powder. Powder samples of the leaves (50 g), stems (50 g), and roots (50 g) were subjected to reflux extraction via 70% ethanol (250 mL × 3 times). The combined extract solutions were filtered through filter paper and concentrated under reduced pressure via a rotary vacuum evaporator, yielding a 70% ethanol extract residue for the leaves, stems, and roots. The resulting extract residues were dissolved in 200 mL of warm water and partitioned successively with n-hexane and ethyl acetate to obtain the n-hexane (Hex), ethyl acetate (EtOAc), and aqueous (W) fractions. All the fractionated samples were stored at 4 °C until further use. Detailed information regarding the extract samples and their respective fractions is provided in Supporting Information Table S1.

Media preparation and chemicals

Mueller–Hinton Agar (MHA) and Mueller–Hinton Broth (MHB) were prepared following the manufacturer’s instructions. The media were sterilized at 121 °C for 15 min at 1 atm. Dimethyl sulfoxide (DMSO) 100% was purchased from Merck, and resazurin was prepared by dissolving 0.015 g of resazurin powder in 100 mL of distilled water (Santa Cruz Biotechnology).

Agar diffusion method and zone of inhibition (ZOI) detection

The agar diffusion method was performed according to the standards set by the Clinical and Laboratory Standards Institute (CLSI). The method is based on the ability of the tested substance to inhibit bacterial growth on an agar medium31. The appearance of a clear zone around the agar well indicates antibacterial activity. The antibacterial activity was assessed by measuring the diameter of the inhibition zone via the following formula: ZOI (mm) = D − d, where D = diameter of the inhibition zone (mm) and d = diameter of the agar well (mm). The detailed method was reported in our previous publication30.

Determination of the minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC)

The minimum inhibitory concentration (MIC) is the lowest concentration of an extract that inhibits the visible growth of microorganisms30. In this method, resazurin, a blue dye, acts as an indicator of cell viability. When microorganisms are metabolically active, they reduce resazurin to resorufin, turning the color from blue to pink. The MIC value is determined by observing the lowest concentration of the extracts or compounds where no color change occurs, indicating the inhibition of microbial activity. The minimum bactericidal concentration (MBC) is the lowest concentration of an extract that kills at least 99.9% of the bacterial population. This was determined by plating bacterial cultures treated with varying concentrations (equal to or greater than the MIC values) of the extract onto Mueller‒Hinton agar (MHA) and observing the growth of colonies. The experiments were repeated twice.

Chemical data of compound preparation

A data mining method from the PubMed database (https://pubmed.ncbi.nlm.nih.gov/) was used to build a dataset of compounds extracted from J. podagrica. A list of 43 compounds from J. podagrica was collected for the molecular docking process (see supporting information). The 3D structures of the compounds were obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov/) and prepared with MarvinSketch (ChemAxon, Cambridge, MA, USA).

Protein preparation

The crystal structure of sortase A (PDB ID: 1T2P) was obtained from the Research Collaboratory for Structural Bioinformatics Protein Data Bank. Polar hydrogen and Kollman charges were added to the protein via AutoDock tools (v. 1.5.6). Finally, the protein was converted to pdbqt format for molecular docking analysis.

Molecular docking study

The 43 selected compounds were screened virtually on the SrtA protein via AutoDock Vina 1.1.2. The grid box covering the protein’s active site was selected from the interactive site of the crystal ligand. Docking scores, reported in kcal/mol, were used to rank the compounds. Finally, the molecular interactions between proteins and selected ligands were visualized via Discovery Studio Visualizer 2020, and 3D and 2D interaction plots were derived30,31.

Molecular dynamics (MD) simulations

MD simulations were performed via the Desmond package (Schrödinger 2020-1, New York, NY, USA)32. The protein-ligand complex was prepared in a 10.0 × 10.0 × 10.0 Å orthorhombic box (simple point‒charge solvation model). Next, a 0.15 M NaCl solution and counterions were added to the system for neutralization. The solvated system was energy-minimized, and its position was restrained with the OPLS3e force field. The minimized system was implemented in an NPT ensemble at 300 K and 1 atm. Finally, the MD simulation was conducted for 100 ns, and 1000 frames were generated, with a recording interval of 100 ps32,33,34.

Fig. 1
figure 1

Agar disc diffusion results of S. aureus (upper) and MRSA (lower) strains extracted from the root, stem, and leaf parts of J. podagrica. (A1, A2): Crude ethanolic extract; (B1, B2): n-hexane fraction; (C1, C2): ethyl acetate fraction; (R – Root extract; S – Stem extract; L – Leaf extract; (-) – Negative control, 10% DMSO; and (+) – Positive control, ampicillin (50 µg/ml).

Full size image
Fig. 2
figure 2

MIC values of different fractions of J. podagrica via a resazurin-based turbidometric assay: crude ethanol (EtOH) extract (A), ethyl acetate (EtOAc) fraction (B), and n-hexane (Hex) fraction (C) of S. aureus (left) and MRSA (right). Ampicillin (50 μg/ml) was used as a positive control, and 10% DMSO was used as a negative control. The numbers indicate the final concentration of extract or fraction in each well.

Full size image
Fig. 3
figure 3

Binding diagram of luteolin with1T2P

Full size image
Fig. 4
figure 4

(A) RMSD and (B) RMSF of luteolin

Full size image
Fig. 5
figure 5

Protein and ligand contact histogram (A) and timeline (B) of luteolin and SrtA

Full size image
Fig. 6
figure 6

L-RMSF (A) and L-property (B), P-SSE histogram (C), and P-SSE timeline (D) of luteolin

Full size image
Fig. 7
figure 7

Torsion analysis of luteolin

Full size image
Fig. 8
figure 8

Determination of the MIC and MBC values of luteolin against S. aureus and MRSA.(A) Resazurin-based broth microdilution method to determine the MIC values of luteolin against S. aureus and MRSA. (B) Bacteria were treated with 1×MIC (left), 2×MIC (middle), or 4×MIC (right) concentrations of luteolin overnight at 37°C before being spread on MHA agar. The growth of S. aureus and MRSA was observed after 24 hr of incubation

Full size image

Absorption, distribution, metabolism, excretion, and toxicity predictions

The studied compounds were evaluated for drug-likeness using “Lipinski’s rule of five”35. Pharmacokinetics-related data, ADMET, were predicted using the SwissADME web tool (http://www.swissadme.ch/index.php)36. Acute oral toxicity predictions were performed using the deep learning-based acute oral toxicity (DL-AOT) prediction server (Center of Quantitative Biology and Molecular Design Laboratory, Peking University, China, 2017)37.