Ertapenem Exposure Is Associated with Downregulation of Purine Biosynthesis Proteins and Inhibition of Biofilm Formation in Staphylococcus aureus

Abstract

Staphylococcus aureus is a major cause of both nosocomial and community-acquired infections, imposing a significant burden on healthcare systems owing to its capacity for mature biofilm formation. These biofilms create polymer-based matrices that reduce bacterial vulnerability to immune responses and antimicrobial agents, complicating the treatment of drug-resistant S. aureus strains and underscoring the need for novel antibacterial targets. Here, we provide the first evidence that FGAM synthase (PurQ), an essential enzyme in the de novo purine biosynthesis pathway, is a putative intracellular protein target associated with ertapenem exposure in S. aureus. Our study demonstrates that ertapenem exhibits potent antibacterial activity against S. aureus (MIC = 1 μg/mL) and is a potential inhibitor of biofilm formation. Mechanistically, this antibiofilm activity correlates with selective downregulation of purine biosynthesis proteins. Using LC-MS/MS analysis of biofilm-derived adherent S. aureus cells, we quantified 1,706 proteins and identified 59 proteins with statistically significant expression changes in ertapenem-treated versus untreated S. aureus cells (49 downregulated, 10 upregulated). Proteins associated with purine biosynthesis were predominantly downregulated, particularly FGAM synthase (PurQ) with a −2.29 log₂ fold change. Molecular docking supported by 100 ns molecular dynamics simulations demonstrated stable ertapenem binding to the PurQ active site (docking score: −9.615 kcal/mol) through interactions with key residues: Gly53, Asp54, Tyr55, Gln89, His141, Gly142, Glu143, and Gly144. Collectively, these findings implicate a novel aspect of ertapenem’s activity, connected to the disruption of purine biosynthesis in S. aureus through targeted downregulation of FGAM synthase. This work provides a strong foundation for drug repurposing strategies aimed at targeting purine biosynthesis in drug-resistant bacterial pathogens.

Introduction

Staphylococcus aureus is a lethal human pathogen well recognized for its ability to form complex bacterial biofilms encased within extracellular matrices. Biofilm formation substantially enhances S. aureus resistance, promotes persistent infections, and reduces traditional medicine efficacy. , This organism asymptomatically colonizes 25–30% of healthy adults, predominantly residing on skin, mucous membranes, nostrils, armpits, and perineum. Transmission occurs via shared personal items, food, contaminated surfaces, nasal secretions, and exposed lesions, with most healthcare transmission occurring via direct contact. S. aureus is a leading cause of infectious endocarditis, bacteremia, skin and soft tissue infections, pleuropulmonary infections, and device-related infections. Multidrug-resistant strains, particularly MRSA (methicillin-resistant S. aureus), present significant treatment challenges. MRSA resistance stems from the mec gene within the Staphylococcal chromosomal cassette mec (SCCmec) region, encoding penicillin-binding protein 2a (PBP-2a) with significantly reduced β-lactam affinity. Consequently, PBP-2a continues peptidoglycan synthesis despite the presence of antibiotics, enabling resistant strain proliferation. ,

One such antibiotic of interest is Ertapenem, a β-lactam antibiotic from the carbapenem class. It works broadly against both Gram-positive and Gram-negative bacteria. Its primary mechanism involves peptidoglycan synthesis inhibition through strong PBP binding, resulting in bacterial cell lysis. , However, recent pharmacokinetic studies and bacterial resistance patterns suggest ertapenem possesses additional mechanisms, prompting investigation into potential secondary targets. Despite limited research on ertapenem’s antibiofilm activity against S. aureus, some antibiotics can disrupt bacterial metabolic processes and prevent biofilm formation. ,

The de novo purine biosynthesis pathway represents a metabolically crucial process for bacterial growth and survival, supplying nucleotides for DNA, RNA, and energy molecules like ATP. This ten-step enzymatic pathway begins with 5-phosphoribosyl-1-pyrophosphate formation and culminates in inosine monophosphate (IMP) production, which is the common precursor to all purine nucleotides. While purine biosynthesis importance has been recognized in cancer chemotherapy and antiviral drug development, , its antibacterial target potential remains underexplored, despite purine nucleotides being crucial for bacterial growth and reproduction. Formylglycinamide ribonucleotide amidotransferase (FGAM synthase), encoded by PurQ, catalyzes the fourth pathway step, converting formylglycinamide ribonucleotide (FGAR) to formylglycinamidine ribonucleotide (FGAM), an essential purine ring formation step. − Notably, bacterial FGAM synthase (PurQ) exhibits significant structural and functional differences from those of its mammalian enzymes, making it a plausible candidate for selective antimicrobial targeting. Previous computational studies have identified small molecules capable of FGAM synthase binding, but none examined interactions with clinically approved antibiotics.

In this study, we employed a multidisciplinary approach combining minimum inhibitory concentration (MIC) determination, biofilm inhibition assays, quantitative proteomic profiling (LC-MS/MS) of biofilm-derived adherent S. aureus cells, molecular docking, and molecular dynamics simulations to examine metabolic responses associated with ertapenem treatment in S. aureus. Our findings reveal significant downregulation of PurQ enzyme and other purine biosynthesis-related proteins following ertapenem exposure, alongside stable in silico interactions between ertapenem and the PurQ active site. Together, these results suggest that ertapenem treatment is associated with perturbation of de novo purine biosynthesis pathways, potentially contributing to its antibiofilm activity in addition to its canonical inhibition of PBPs. This work provides mechanistic insights into ertapenem-induced metabolic responses and highlights purine metabolism as a complementary pathway of interest for antibacterial and drug-repurposing strategies against resistant S. aureus infections.

Results

Ertapenem Exhibits Potent Antibiofilm Activity against S. aureus

In this study, the minimum inhibitory concentration (MIC) of ertapenem against S. aureus ATCC 29213 was determined to be 1 μg/mL. The subsequent biofilm formation studies showed remarkable dose-dependent inhibition of biofilm development upon the challenge of bacterial cultures with a range of inhibitory doses (16 μg/mL–0.031 μg/mL) of ertapenem (Figure ). A prominent decrease in biofilm formation can be seen in Figure S1, where at 1/2 the MIC concentration of ertapenem, a visibly faded biofilm (blue ring) has resulted in the test tube. Additionally, without ertapenem, the surface of this rough organic membrane was seen to be completely overlaid with S. aureus microcolonies and aggregates as seen after the SEM analysis (Figure A). However, when sub-MICs of ertapenem were added to the culture media, only limited microcolonies of S. aureus were seen on the rough organic membrane (Figure B and C). As a result, quantitative biofilm biomass analysis showed remarkable reductions in biofilm formation across all ertapenem doses evaluated compared to untreated control groups (p < 0.001), suggesting an apparent dose–response relationship.

1.

Effect of ertapenem on biofilm formation in Staphylococcus aureus ATCC29213. Biofilm biomass was quantified using the crystal violet assay after treatment with increasing ertapenem concentrations (16–0.03125 μg/mL). Optical density measured at 595 nm showed a dose-dependent reduction in biofilm biomass with higher ertapenem concentrations. Data are presented as mean ± standard deviation from three independent experiments performed in triplicate.

2.

Scanning electron micrographs showing ertapenem effects on Staphylococcus aureus biofilm morphology. (A) Untreated control showing dense, multilayered cocci aggregates with intact morphology; (B) ertapenem treatment at 1/4-MIC concentration resulting in disrupted cell aggregation and reduced biofilm density; (C) ertapenem treatment at 1/2-MIC concentration showing severe biofilm disruption with mostly dispersed, damaged cells.

Proteomic Profiling of Ertapenem-Treated S. aureus

To investigate how S. aureus responds to subinhibitory concentrations of ertapenem, we carried out detailed label-free quantitative proteomics of biofilm-derived adherent S. aureus cells using high-resolution liquid chromatography-tandem mass spectrometry (LC-MS/MS). After processing the data and filling in missing values by using the k-nearest neighbor algorithm, we identified and quantified a total of 1,706 proteins in both ertapenem-treated (1/2-MIC, 0.5 μg/mL) and control samples. The strict statistical filtering criteria (fold change ≥ 2.0 or ≤ 0.5) and a significance threshold (p < 0.05) find 59 proteins that were significantly differentially expressed. The proteomic response showed mostly downregulation, with 49 proteins downregulated compared to 10 proteins that were upregulated relative to untreated controls (Table S1). Differentially expressed proteins were identified using complementary statistical methods to ensure reliable detection of biologically important changes. Volcano plot analysis (fold change ≥ 2.0, p < 0.05) provided a clear view of statistical significance versus biological relevance (Figure ). Principal component analysis (PCA) and partial least-squares-discriminant analysis (PLS-DA) helped identify the two distinct groups. The most significantly downregulated proteins included aspartate 1-decarboxylase, phosphoribosylformylglycinamidine synthase subunit PurQ, carbamate kinase 1, ribonuclease HIII, protein GrpE, lactonase drp35, and global transcriptional regulator Spx (Table ).

3.

Volcano plot of statistically significantly differentially expressed proteins selected with a fold change threshold of 2.0 and t-test threshold (p) of 0.05.

1. List of the Top Differentially Expressed Proteins in the Ertapenem-Treated Test Sample (FC ≥ 2, P-Value ≤ 0.05).

Accession o.	Protein name	Fold change (treated/control)	Log2(FC)	t-test (P)
Downregulated in Ertapenem-Treated Cells
Q6GI16	Phosphoribosylformylglycinamidine synthase subunit PurQ	0.20454	–2.2896	0.015136
Q8NX59	Ribonuclease HIII	0.11744	–3.09	0.003391
Q2YX63	UPF0358 protein SAB0977	0.15031	–2.734	0.02619
Q6GAE5	Phosphoribosylformylglycinamidine synthase subunit PurQ	0.27831	–1.8452	0.015937
Q2YWG1	Aspartate 1-decarboxylase	0.28936	–1.789	0.021897
Q2FHR7	Carbamate kinase 1	0.36122	–1.4691	0.041959
Q2FXZ1	Protein GrpE	0.41333	–1.2746	0.024197
Q8NUH4	Lactonase drp35	0.44374	–1.1722	0.003138
A7X203	Nuclease SbcCD subunit C	0.47512	–1.0736	0.000753
P60380	Global transcriptional regulator Spx	0.49094	–1.0264	0.000341
Upregulated in Ertapenem-Treated Cells
P07944	Beta-lactam-inducible penicillin-binding protein	11.855	3.5674	2.83 × 10–5
Q5HH83	ClpXP adapter protein SpxH	2.4492	1.2923	0.002385
Q2YT45	Heat-inducible transcription repressor HrcA	2.1228	1.086	0.005899

After analyzing the complete quantified proteome, which included 1,706 proteins, a clear distinction was revealed between treated and untreated groups. This confirms that ertapenem creates a strong and unique proteomic signature. Hierarchical clustering analysis of differentially expressed proteins (DEPs) provided further support for treatment-specific expression patterns, showing coordinated changes in related protein networks. Given the observed inhibitory effects of ertapenem on S. aureus biofilm formation, special focus was given to proteins involved in biofilm processes. Among these, the purine biosynthesis enzyme PurQ showed significant downregulation of ∼4.9-fold. This suggests that changes in nucleotide synthesis pathways may play a role in the antibiofilm effects of subinhibitory concentrations of ertapenem.

Principal Component Analysis (PCA), Partial Least-Squares-Discriminant Analysis (PLS-DA), and Hierarchical Clustering Analysis

Principal component analysis (PCA) was used to evaluate overall variance patterns in the proteomic data set and the discriminatory power of differentially expressed proteins. The data variance was expressed as scores, which were plotted to reveal the differences between the two groups. Notably, PCA score plots indicated distinct separation between control and ertapenem-treated groups, indicative of distinctive proteomic signatures associated with subinhibitory ertapenem exposure (Figure A). To confirm these results and assess the supervised classification potency, partial least-squares-discriminant analysis (PLS-DA) was carried out on the same differentially expressed protein data set. PLS-DA modeling corroborated the result of PCA, with distinct intergroup separation and high predictive capability, independently confirming the biological significance of observed proteomic changes (Figure B).

4.

(A) Principal Component Analysis (PCA) score plot showing clear separation between ertapenem-treated (test) and control groups based on significantly differentially expressed proteins (DEPs). The first two principal components account for 89.2% of total variance (PC1:75.4%, PC2:13.8%). (B) Partial least-squares-discriminant analysis (PLS-DA) score plot showing distinct test and control group clustering. Component 1 explains 73.5%, and Component 2 explains 15.5% of the variance, confirming clear group separation.

Hierarchical cluster analysis (HCA) was conducted to identify proteins with similar expression patterns and validate the clustering behavior of samples based on treatment conditions. The analysis of the complete proteomic data set, which includes 1,706 proteins, showed clear separation between the ertapenem-treated and control samples. All biological replicates were grouped on the basis of their treatment without any mix-ups. This unsupervised clustering method highlights the consistency and reliability of the proteomic response to ertapenem treatment. Here, we have provided the clustering analysis for the top 100 proteins for clearer visualization (Figure ).

5.

Hierarchical clustering heatmap of significantly differentially expressed proteins (DEPs) between ertapenem-treated (T1–T3) and control (C1–C3) groups. Rows represent individual proteins, and columns represent biological replicates. Expression levels are color-coded: red indicates upregulated, and blue indicates downregulated. Clustering shows samples within groups cluster together, highlighting clear protein expression profile differences between control and ertapenem-treated groups.

Overall, these multivariate statistical analyses show the strong ability of the identified differentially expressed proteins to distinguish between the groups and confirm the statistical importance of the molecular responses to subinhibitory ertapenem exposure.

Significant Downregulation of FGAM Synthase (PurQ)

Among the most significantly downregulated proteins, Phosphoribosylformylglycinamidine synthase subunit PurQ (UniProt ID: Q6GI16), a key de novo purine biosynthesis enzyme, revealed a −2.29 log₂ fold change, equivalent to ∼4.9-fold reduced protein levels versus untreated controls. PurQ functions as formylglycinamidine ribonucleotide (FGAM) synthase, catalyzing ATP-dependent formylglycinamide ribonucleotide (FGAR) to formylglycinamidine ribonucleotide (FGAM) conversion, the fourth enzymatic step in purine ring biosynthesis. This reaction represents one of the most energetically costly steps in purine nucleotide biosynthesis, involving both ATP hydrolysis and glutamine amidation. Strong PurQ expression suppression implies core metabolic reorganization away from nucleotide biosynthesis, potentially representing adaptive responses to ertapenem-induced cellular stress.

KEGG pathway enrichment analysis mapped PurQ onto the S. aureus purine biosynthetic network, revealing complex interaction networks among the identified proteins, associated pathways, and cellular metabolic processes. A total of 15 pathways were significantly impacted, including amino acid biosynthesis, glyoxylate metabolism, folate biosynthesis, riboflavin metabolism, thiamine metabolism, pyruvate metabolism, and histidine metabolism (Figure S2). Our study highlights purine biosynthesis as a key contributor to biofilm formation with FGAM synthase (PurQ) playing a crucial role in S. aureus biofilm formation.

Molecular Docking Analysis between PurQ and Ertapenem

Since PurQ plays a critical role in S. aureus biofilm formation and had a significant reduction after subinhibitory ertapenem treatment (1/2-MIC, 0.5 μg/mL), molecular docking studies were employed to examine potential ertapenem binding interactions with the PurQ enzyme. Docking was carried out at two potential binding sites (Site-1 and Site-2) within PurQ to assess preferential ligand recognition. The docking scores of ertapenem were −9.615 kcal/mol for Site-1 and −6.732 kcal/mol for Site-2, indicating favorable binding affinities at both sites. At Site-1, molecular docking revealed a preferred binding conformation of ertapenem within the catalytic pocket, characterized by the formation of six hydrogen bonds with key residues. Specifically, ertapenem formed interactions with Gly53, Asp54, Tyr55, Gln89, His141, Gly142, Glu143, and Gly144, all of which reside in the binding cavity of the enzyme (Figure ).

6.

Molecular docking interaction of FGAR (control) and ertapenem within the PurQ enzyme active site pocket. Protein surface represented in yellow with electrostatic potential mapping highlighting polar regions (blue: positive, red: negative). (A) 3D interaction of FGAR is shown in orange stick representation with interacting residues in cyan. (B) 3D interaction of ertapenem is shown in magenta stick representation. Black dashed lines indicate hydrogen bonding interactions stabilizing ligand–protein complexes. (C) 2D interaction diagram of FGAR with PurQ enzyme. (D) 2D interaction diagram of ertapenem with PurQ enzyme.

At site 2, docking also revealed stable binding conformations, though with relatively lower affinity compared to site 1. The predicted binding mode suggests stabilizing interactions; however, the overall binding energy and residue-level contacts were less significant than those observed at site 1. This indicates that ertapenem binding could interfere with normal substrate recognition and enzymatic turnover.

Molecular Dynamics Simulations

To validate the molecular docking predictions and to assess the stability of the ertapenem-PurQ complex over extended periods of time, comprehensive molecular dynamics (MD) simulations were performed along a 100 ns trajectory. We visualized the simulation trajectories to assess whether Site-1 (positioned between two flexible loops) or Site-2 (situated between two α-helices) exhibited better ligand retention. An overlay of ligand positions, sampled every 10 ns throughout the simulation, onto the average protein structure is presented in Figure . At Site-1, both moleculescontrol (FGAR; Figure A) and ertapenem (Figure B)remained stably bound, showing only minor fluctuations during the course of the simulation. In contrast, at Site-2, both FGAR (Figure C) and ertapenem (Figure D) displayed considerable structural fluctuations and were not efficiently retained within the binding pocket. Taken together, these observations suggest that Site-1 provides a more stable environment for ligand binding and may represent the potential interaction site.

7.

Overlay of ligand trajectories sampled every 10 ns onto the average protein structure, illustrating ligand retention at two proposed binding sites. (A) Control molecule FGAR at Site-1 (between flexible loops) and (B) ertapenem at Site-1 show stable binding with minimal fluctuations. (C) FGAR at Site-2 (between α-helices) and (D) ertapenem at Site-2 exhibit significant structural fluctuations and poor pocket retention. Results indicate Site-1 provides a more favorable ligand-binding environment.

We analyzed the structural stability of the protein–ligand complexes by plotting the RMSD (root-mean-square deviation) values as a function of simulation time (Figure A). Consistent with the visual inspection of trajectories, Systems A and B exhibited stable RMSD profiles with fluctuations remaining within ∼2 Å throughout the simulation, indicating well-retained ligands at Site-1. In contrast, Systems C and D showed markedly higher RMSD fluctuations, reflecting reduced stability of the complexes. This instability can be attributed to the weaker retention of ligands at Site-2, leading to greater conformational flexibility of the protein–ligand complexes compared with those bound at Site-1. To further validate these observations, we examined the radius of gyration (RoG) values (Figure B). Systems C and D displayed broader RoG fluctuations and a more expanded structural profile compared to Systems A and B, reinforcing the conclusion that ligand binding at Site-2 is less stable. Together, these results suggest that Site-1 provides a more favorable and stable binding environment, while Site-2 is characterized by weaker ligand interactions and higher structural variability.

8.

(A) Root mean square deviation (RMSD) profiles of protein–ligand complexes as functions of simulation time. Systems A and B (Site-1 ligands) exhibit stable fluctuations within ∼2 Å, indicating well-retained ligand binding. Systems C and D (Site-2 ligands) show larger deviations, consistent with weaker ligand retention and reduced complex stability. (B) Radius of gyration (RoG) plots for protein–ligand complexes. Systems A and B display compact and stable structural profiles, whereas Systems C and D show broader RoG fluctuations and more expanded conformations, suggesting less stable Site-2 binding.

The binding affinities of the protein–ligand complexes were estimated using the MM-PBSA approach over the last 50 ns of the trajectories (200 frames). The computed binding free energies for systems A, B, C, and D were −8.73 kcal/mol, −12.47 kcal/mol, −4.82 kcal/mol, and −4.61 kcal/mol, respectively (Table ).

2. Comparative Docking Scores of Ertapenem and FGAR with PurQ Enzyme.

Ligands	Binding site	Docking score (kcal/mol)	Glide energy (kcal/mol)	Binding free energy (kcal/mol)
Ertapenem	Site-1	–9.615	–67.339	–12.47
Ertapenem	Site-2	–6.732	–58.361	–4.61
FGAR (control)	Site-1	–7.424	–61.761	–8.73
FGAR (control)	Site-2	–5.911	–56.689	–4.82

Notably, the ligands bound at site 2 (systems C and D) exhibited significantly weaker binding affinities compared to those at Site-1 (systems A and B). Furthermore, at Site-1, the candidate ligand ertapenem (system B) showed considerably stronger binding affinity than the control ligand FGAR (system A). To further rationalize the higher binding observed in system B, representative structures from the MD simulations were used to generate 2D interaction plots (Figure ).

9.

Two-dimensional interaction plots of ligand binding at Site-1. (A) Control ligand FGAR (System A) and (B) candidate ligand ertapenem (System B) both form three hydrogen bonds at the active sites. However, ertapenem exhibits denser van der Waals interaction networks and adopts snug-fit conformations compared to FGAR, consistent with higher binding affinity.

Both systems A and B formed three hydrogen bonds at the active site; however, system B displayed a denser van der Waals interaction network and adopted a snug-fit conformation within the binding pocket. These additional stabilizing interactions likely account for its superior binding affinity relative to the control. Overall, our results indicate that ertapenem at site-1 binds more strongly than FGAR, suggesting that ertapenem may serve as a promising lead molecule for targeting the PurQ enzyme. The predicted binding affinity and interaction patterns support the hypothesis that ertapenem may engage intracellular metabolic enzymes in addition to its established penicillin-binding protein targets, potentially contributing to pleiotropic cellular effects under antibiotic stress.

Discussion

This study provides evidence supporting an association between ertapenem exposure and perturbation of purine biosynthesis, including downregulation of FGAM synthase (PurQ) in S. aureus. Although ertapenem is conventionally recognized as a β-lactam antibiotic inhibiting penicillin-binding proteins in bacterial cell wall synthesis, our findings suggest that ertapenem treatment is associated with alterations in intracellular metabolic pathways that may contribute to its antibacterial and antibiofilm effects. Importantly, we do not propose PurQ as an alternative primary target; rather, it is presented as a putative secondary pathway affected under ertapenem-induced stress conditions. Biofilm inhibition assays provide functional phenotypic evidence consistent with these molecular and metabolic perturbations, although direct causality between purine biosynthesis disruption and biofilm inhibition cannot be established from the current data.

Dose-dependent reductions in biofilm formation at sub-MIC concentrations indicate that metabolic reprogramming, potentially including partial disruption of purine biosynthesis, can significantly influence bacterial physiology and virulence-associated behaviors. Such effects are particularly relevant in the context of biofilm-associated infections, which are difficult to eradicate and represent a major driver of antimicrobial resistance. Nevertheless, it must be acknowledged that biofilm inhibition at sub-MIC levels may also arise from global stress responses, altered growth dynamics, or broader metabolic shifts induced by antibiotic exposure. Furthermore, this study was conducted using a single reference strain (S. aureus ATCC 29213), which may not fully capture the phenotypic, metabolic, and proteomic diversity observed across all of the clinical isolates.

Recent studies provide contextual support for the involvement of purine metabolism in biofilm regulation, strengthening the biological plausibility of our observations. De novo purine biosynthesis pathways have been reported to be upregulated during biofilm formation in both Staphylococcus aureus and Enterococcus faecalis. Disrupting this pathway, either genetically or pharmacologically, significantly impairs biofilm development. Moreover, mechanistic studies show purine biosynthesis connects with signaling molecule c-di-AMP production, influencing extracellular DNA release and biofilm persistence in MRSA. While these studies support a role for purine biosynthesis in biofilm formation, their relevance to ertapenem’s mode of action remains indirect and associative. Another study reported that antifolates like trimethoprim–sulfamethoxazole represent one of the few truly synergistic antibacterial combinations and remain the only clinically used antifolate therapy for bacterial infections in metabolic pathways. Despite emerging resistance, it continues to serve as a key first-line treatment, highlighting the need for next-generation antifolates to sustain clinical efficacy.

The proteomic analysis of biofilm-derived adherent S. aureus cells further supports the selective nature of ertapenem-associated metabolic responses. Among 1,706 LC-MS/MS-identified proteins, only 59 were significantly differentially expressed in ertapenem-treated cells. The ∼4.9-fold downregulation of PurQ highlights purine biosynthesis as a prominently affected pathway; however, pathway enrichment also revealed alterations in amino acid biosynthesis, central carbon metabolism, and peptidoglycan synthesis processes, suggesting that purine deprivation may contribute to broader downstream metabolic effects rather than acting in isolation. Computational analysis demonstrated strong ertapenem binding to PurQ compared to natural substrate FGAR. Molecular dynamics simulations confirmed stable binding over 100 ns time scales. Key active site residue involvement (Gly53, Asp54, Tyr55, Gln89, His141, Gly142, Glu143, and Gly144) in hydrogen bonding offers molecular explanations for observed inhibition. These findings are particularly important given PurQ’s vital role in de novo purine biosynthesis pathways, providing essential nucleotides for DNA, RNA, and ATP synthesis.

From a clinical perspective, these findings offer additional insights into the complex cellular responses elicited by ertapenem treatment. The observed metabolic perturbations may help explain antibacterial and antibiofilm effects that cannot be fully explained by classical β-lactam mechanisms alone. Moreover, it suggests that combining treatments targeting cell wall synthesis and purine metabolism could improve antibacterial effects. Overall, these findings highlight that enzymes involved in purine biosynthesis represent a potentially overlooked and therapeutically relevant target for the development of new antibacterial strategies. Importantly, we do not suggest that purine biosynthesis inhibition overcomes established β-lactam resistance mechanisms, but rather that it may represent a complementary vulnerability under antibiotic stress. However, several limitations require consideration. Only one bacterial strain and antibiotic concentration were evaluated; broader verification across various S. aureus strains, including clinical MRSA isolates, would strengthen these conclusions. Future research should focus on studying this mechanism in other bacterial species, targeted transcriptional validation (including qRT-PCR analysis of purQ and related purine biosynthesis genes), optimizing ertapenem analogs for improved purine biosynthesis inhibition, evaluating MBIC and mature biofilm eradication, and exploring combination therapies targeting both cell wall synthesis and nucleotide metabolism simultaneously.

Conclusions

This study suggests a potential association between FGAM synthase (PurQ) and ertapenem exposure in S. aureus. Integrated proteomic and computational analyses revealed a correlation among ertapenem treatment, downregulation of PurQ, and alterations in protein expression involved in the purine biosynthesis pathway. Ertapenem exposure at sub-MIC concentrations is associated with reduced biofilm formation and decreased purine biosynthetic enzyme levels, suggesting a potential clinical relevance. Collectively, these findings highlight a previously underappreciated metabolic response to ertapenem beyond its canonical cell-wall-targeting activity. The results underscore the potential relevance of purine metabolism as a complementary pathway influencing biofilm formation and the antibiotic response in S. aureus. This work provides a hypothesis-generating framework for future studies aimed at exploring metabolic vulnerabilities that could be leveraged to enhance antibacterial and antibiofilm strategies.

Materials and Methods

Bacterial Strain and Growth Conditions

The bacterial strains were obtained from the National Collection of Industrial Microorganisms (NCIM, Pune, India). S. aureus was grown at 37 °C in Mueller-Hinton Broth (MHB) for testing its susceptibility to ertapenem. Kanamycin was purchased from Sigma-Aldrich (India). All the media components were purchased from Himedia (India).

Determination of Minimum Inhibitory Concentration of Ertapenem

The minimum inhibitory concentration (MIC) of ertapenem (Zydus Healthcare Ltd., Ahmedabad, India) was determined through the standard broth microdilution method following Clinical and Laboratory Standards Institute (CLSI) guidelines M07-A10. 2-fold serial dilutions of ertapenem were done in Mueller–Hinton broth, followed by inoculation with standard bacterial suspensions of S. aureus (final inoculum ≈ 5 × 10⁵ CFU/mL). Appropriate controls were untreated wild-type S. aureus ATCC29213 as a negative control and sterile culture medium as a vehicle control. After 24 h of incubation at 37 °C, optical density was measured at 600 nm through a microplate reader (BioTek Cytation, Agilent Technologies, Santa Clara, CA). The experiments were done in triplicate to ensure reproducibility and statistical validity.

Inhibition of Biofilm Formation by Ertapenem Using Tissue Culture Plate (TCP) Assay

The overnight culture of S. aureus ATCC29213 at the midexponential phase of growth was diluted with sterile saline to make a 0.5 McFarland standard, which was further diluted 1:100 in MH media to get a final 1–2 × 10⁶ CFU/mL in a microtiter plate. Ertapenem is serially diluted in the microtiter plate to get 10 serial dilutions (16 μg/mL to 0.031 μg/mL). Then, 100 μL of prepared bacterial suspension was added to each well. In addition, a control (only MH broth) and a negative control (only bacterial cells) were included and incubated at 37 °C for 72 h. The supernatant was then removed, the wells were rinsed three times with phosphate-buffered saline (PBS; pH 7.4), 200 μL of 99% methanol was added to the wells to fix the biofilms, and then, the plates were decanted after 15 min and stained for 5 min with 200 μL of 2% crystal violet. The wells were rinsed with PBS (pH 7.4), and the dye was resolubilized with 200 μL of 33% (v/v) glacial acetic acid per well. All wells were then measured using a Biotek Cytation 5 Imaging Multimode Reader (Agilent, Santa Clara, CA) at 595 nm. The experiments were performed in triplicate. A similar experiment was done by the macrodilution method in a sterile test tube at 1/2 MIC (0.5 μg/mL) in 5 mL of MH broth containing ertapenem. The result was interpreted visually.

Scanning Electron Microscopy (SEM)

Antibiofilm effects were examined using SEM at 1/2 MIC and 1/4 MIC ertapenem concentrations. Bacterial suspensions (1 mL) supplemented with appropriate drug amounts were transferred to separate microcentrifuge tubes, with untreated controls. After 72 h incubation at 37 °C, planktonic cells were removed, and biofilm layers collected, washed three times with PBS, fixed in 4% Karnovsky’s fixative for 4 h, and air-dried. Samples were gold-coated and observed with a Zeiss Evo 18 scanning electron microscope (Sophisticated Electron Microscopy Unit, AIIMS, New Delhi). Two images from three independent replicates were captured and analyzed.

Preparation of Protein Extracts

Biofilm-derived adherent S. aureus cells (ATCC29213), supplemented with 1/2-MIC (0.5 μg/mL) ertapenem, were grown in MHB media in 96-well plates at 37 °C for 24 h, while control samples remained untreated. After incubation, the supernatants were removed, and the wells were rinsed twice using sterile PBS buffer (50 mM, pH 7.4). We harvested the adherent biofilms by scraping them mechanically and then dissolved them in lysis buffer for protein extraction. All experiments were done in triplicate, with untreated cells used as controls.

Liquid Chromatography–Mass Spectrometry (LC-MS/MS) Analysis

Sample Preparation

Protein samples were subjected to tryptic digestion following standard protocols. Briefly, proteins were reduced with 5 mM tris­(2-carboxyethyl)­phosphine (TCEP) and alkylated with 50 mM iodoacetamide to prevent disulfide bond reformation. Enzymatic digestion was performed using sequencing-grade trypsin at a 1:50 trypsin/lysate ratio for 16 h at 37 °C. Resulting peptide mixtures were desalted using C18 silica cartridges to remove salts and contaminants, followed by vacuum centrifugation to dryness. The dried pellet was resuspended in buffer A (2% acetonitrile, 0.1% formic acid) before LC-MS/MS analysis.

Mass Spectrometric Analysis

The experiments used an Easy-nLC 1000 ultrahigh-pressure liquid chromatography system connected to an Orbitrap Exploris mass spectrometer (Thermo Fisher Scientific). Peptide samples (1 μg) were loaded onto a C18 analytical column (15 cm × 75 μm, 3.0 μm Acclaim PepMap, Thermo Fisher Scientific) and separated using a linear gradient of 0–40% buffer B (80% acetonitrile, 0.1% formic acid) at a flow rate of 500 nL/min. The samples were then injected for MS analysis. The LC gradients were run for 110 min. Full-scan MS1 spectra were acquired in the Orbitrap analyzer (Max IT = 60 ms, AGC target = 300%, RF Lens = 70%, R = 60,000, mass range = 375–1500; profile data). The top 20 most abundant precursor ions from MS2 spectra were selected for higher-energy collisional dissociation (HCD) fragmentation. MS2 spectra were acquired at 15,000 resolution with an AGC target of 100% and a maximum injection time of 60 ms. Dynamic exclusion was employed for 30 s, excluding all charge states for a given precursor.

Data Processing

All samples were processed and generated raw files were analyzed with Proteome Discoverer (v2.5) against the Uniprot Staphylococcus aureus database. For dual Sequest and Amanda searches, the precursor and fragment mass tolerances were set at 10 ppm and 0.02 Da, respectively. The protease used to generate peptides, i.e., enzyme specificity, was set for trypsin/P (cleavage at the C terminus of “K/R: unless followed by “P”). Carbamidomethylation on cysteine as a fixed modification and oxidation of methionine and N-terminal acetylation were considered as variable modifications for the database search. Both peptide spectrum match and protein false discovery rates were set to 0.01 FDR.

Statistical Analysis

All experiments were performed in triplicate. Protein abundance data underwent quality control and preprocessing via MetaboAnalyst 6.0. Missing values were imputed using k-nearest neighbor algorithms, and data were normalized via Pareto scaling. Differentially expressed proteins (DEPs) were detected using unpaired t-tests with p ≤ 0.05 significance. Volcano plots illustrated statistical and biological significance. Principal component analysis (PCA) determined data variance and sample clustering, while partial least-squares-discriminant analysis (PLS-DA) optimized group separation. Hierarchical clustering analysis illustrated protein expression patterns. Results are expressed as mean ± standard deviation from three independent replicates.

Computational Studies

Protein Structure Preparation

Due to the absence of an experimentally determined Staphylococcus aureus FGAM synthase (PurQ) crystal structure, the modeled 3D structure was downloaded from AlphaFold Protein Structure Database (AF-ID: AF-Q6GI16-F1-v4; confidence score pLDDT > 90). Structure quality was evaluated using the PROCHECK server for stereochemical integrity, Ramachandran plots, bond lengths, angles, and peptide bond planarity. Active-site residues were inferred from conserved catalytic motifs using the SiteMap module within Schrödinger’s Maestro Suite. SiteMap analysis identified four probable binding sites ranked by SiteScore (site enclosure and size) and Dscore (druggability assessment). The highest-scoring cavity (SiteScore: 0.715, Dscore: 0.698) Site-1 was selected for molecular docking. Site-2 was also considered due to similar scores (SiteScore: 0.712, Dscore: 0.651). Protein structures were preprocessed through Protein Preparation Wizard with bond order assignment, hydrogen bonding network optimization, water molecule deletion beyond 5 Å, and energy minimization using the OPLS3 force field.

Ligand Preparation

The 3D structure of ertapenem (PubChem CID: 150610) and FGAR (native substrate of PurQ) (PubChem CID: 130805) was retrieved from the PubChem database and underwent extensive preparation via the LigPrep module of Schrödinger Suite 2017-1. For biologically relevant conformations, physiologically appropriate protonation states were computed at pH 7.0 ± 0.5 via the Epik ionization state prediction algorithm, which uses quantum mechanical computations to predict pK _a values with good accuracy. Tautomeric and stereoisomeric forms of all possible types were enumeratively treated and saved for further analysis to account for the complete conformational variability of ertapenem in a purely physiological environment. Energy minimization was carried out via the OPLS3 force field with 0.05 kJ/mol/Å as a setting for the maximum derivative of the convergence criteria. Final LigPrep-processed ligand structures were transformed to Maestro format for molecular docking, where rotatable bonds were identified automatically and an appropriate number of torsional degrees of freedom were assigned to facilitate exhaustive conformational sampling during docking.

Molecular Docking between PurQ and Ertapenem

Molecular docking was performed using the Glide module (Schrödinger Release 2017-1) in Standard and Extra Precision modes. 3D FGAM synthase (PurQ) structures were prepared with 20 × 20 × 20 Å receptor grid boxes generated around predicted binding sites. Ligands, including ertapenem and FGAR (control), were docked using flexible sampling with van der Waals radii scaled to 0.8 and a partial charge cutoff of 0.15. Multiple docking runs with different random seeds ensured reproducibility. Top-ranking poses were selected based on interaction profiles, docking scores, and Glide energies. Final docked complexes were analyzed through PyMOL v2.3.

Molecular Dynamics Simulations

We performed a detailed investigation of the time-dependent binding of ligands to the target protein using all-atom explicit-solvent molecular dynamics (MD) simulations. Over the past decade, MD has emerged as a powerful tool to unravel fundamental biological phenomena, including protein/nucleic acid-ligand recognition, solvent-mediated effects, and macromolecular interactions. − For our study, the docked protein–ligand complexes served as the starting coordinates for simulations, which were executed using the AMBER software suite. Since the precise ligand-binding site on the protein was unknown, two prominent candidate pockets were explored (generated in the docking step). Site-1 was located between two flexible loops, while Site-2 was situated between two α-helices. The protein was parametrized with the modified ff99SB force field, while counterions (added to preserve system electroneutrality) were modeled with Joung and Cheatham parameters. Ligand parameters were derived from geometry-optimized structures obtained via Gaussian09 using the HF/6-31G* basis set, , and an AMBER-compatible library (.lib) file was generated, encoding all bond, angle, dihedral, and charge details. Each complex was solvated in a cubic TIP3P water box with counterions, employing periodic boundary conditions. Particle Mesh Ewald (PME) was used for long-range electrostatics, and hydrogen bond constraints were enforced using the SHAKE algorithm. The resulting topology (.top) and coordinate (.crd) files were used to build the simulation setup. Energy minimization proceeded in stages, initially relaxing solvent and ions with the protein–ligand complex restrained, followed by unrestrained minimization of the full system. The equilibrated systems were gradually heated to room temperature (50 ps, NVT) and equilibrated further for 20 ns. Subsequently, 100 ns production runs were conducted for each system. Simulation trajectories were analyzed with CPPTRAJ to compute RMSD and radius of gyration (RoG), and to generate overlay images for examining structural fluctuations. Binding energetics were estimated via MM-PBSA, , while LigPlot+ was employed to generate 2D interaction maps of protein–ligand contacts.

Glossary

Abbreviations

DEPs

Differentially Expressed Proteins

FGAR

Formylglycinamide Ribonucleotide

FGAM

Formylglycinamidine Ribonucleotide

LC-MS/MS

Liquid Chromatography-Tandem Mass Spectrometry

MD

Molecular Dynamics

MIC

Minimum Inhibitory Concentration

MRSA

Methicillin-Resistant Staphylococcus aureus

PCA

Principal Component Analysis

PLSDA

Partial Least-Square-Discriminant Analysis

RMSD

Root Mean Square Deviation

RMSF

Root Mean Square Fluctuation

The mass spectrometry proteomics data and Table S1 have been deposited in a publicly accessible GitHub repository at https://github.com/AnamikaSingh08/Dr.-Pradeep-Sharma_Lab.git. The raw LC-MS/MS proteomics data are also being prepared for deposition to the PRIDE repository.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c11260.

• Additional experimental details including, figures of experimental results (PDF)

A.S. conducted experiments, acquired data, compiled results and methods, and wrote the manuscript. S.P. conducted antibacterial studies, P.P. conducted MD simulations, T.P.S., S.S., and P.S. acquired funding, analyzed data, and supervised. All authors have reviewed, critiqued, and approved the final manuscript.

The authors declare no competing financial interest.