Identification of Bioactive Postbiotics Against Neonatal Meningitis Caused by Group B Streptococcus via Srr2-Targeted In silico Screening

1. INTRODUCTION

Neonatal bacterial meningitis remains a life-threatening condition with substantial global morbidity and mortality. Immature immunity and limited physiologic responses increase the impact of CNS infection. Therefore, it increases the need for prompt recognition and treatment in the first 28 days of life [1, 2]. Clinically, early-onset disease typically presents within the first week of life. However, late-onset disease emerges subsequently in the neonatal period; both forms demand urgent evaluation because delayed therapy degenerates neurological outcomes [1, 2]. Group B Streptococcus (GBS) remains a leading cause of neonatal meningitis globally. It encapsulates a bacterium that commonly settles in the gastrointestinal and genital tracts of more than 50% of adult individuals [3]. The transmission is most often tied to maternal gastrointestinal or genital colonisation and exposure during labour and delivery [3, 4]. After colonisation, bacteria access the bloodstream, cross protective barriers to seed the meninges. It causes irritability, fever, feeding difficulty, respiratory distress, or bulging fontanelle [5].

Although the efficacy of antimicrobial agents and probiotics against viral colonisation of the neonatal gut has been studied widely, none of the studies has been specific to virulence factors like the Srr2 protein in Group B Streptococcus (GBS). Such research tends to concentrate on broad-spectrum antibiotics or even general antimicrobial action without much consideration of antivirulence of postbiotic metabolites [6, 7]. Unlike previous research, which aims at broad-based pathogen inhibition, the study employs a directed antivulence approach, which prevents bacterial adhesion [8]. This would reduce the risk of GBS-induced neonatal meningitis but spare the normal neonatal microbiome. The present results require orthogonal computational validation (cross-engine redocking, replicate MD, MM-GBSA/MM-PBSA with residue-wise decomposition), biophysical confirmation of target engagement, and experimental assays in CC17 clinical isolates assessing adhesion to brain endothelial cells, biofilm formation, and synergy with first-line antibiotics [9]. Neonatal pharmacology is characterized by the developmental status of the organs, particularly the blood-brain barrier (BBB) and immature metabolic pathways that have a significant impact on drug absorption, distribution, metabolism, and excretion (ADME) properties. Factors, including interactions with breast milk or the infant formula, can influence the exposure and efficacy of postbiotics.

Against this backdrop, postbiotics and postbiotic-derived metabolites have attracted attention as adjunct or alternative strategies to antibiotic therapy. Postbiotics, defined by ISAPP as preparations of inanimate microorganisms and/or their components that confer health benefits, represent a broader category under which microbially derived components (including cell-wall fragments, peptides) may act directly on host or microbe [10]. A targeted antivirulence strategy was pursued for GBS centred on serine-rich repeat protein 2 (Srr2), a surface adhesin enriched in hypervirulent CC17 lineages that dominate neonatal invasive disease [8]. Srr2 contributes to adhesion and traversal of host barriers, including interactions with fibrinogen and other host ligands, and forms part of the molecular toolkit that enables CC17 lineages to invade brain endothelium [8]. However, the past studies did not focus on bioactive postbiotics against neonatal meningitis. By focusing on adhesion rather than bacterial viability, antivirulence approaches seek to blunt pathogenesis while potentially lowering selection for classical antibiotic resistance, an attractive principle for fragile neonatal hosts.

This study will examine how citric acid, a postbiotic-related metabolite, can serve as an antivirulence agent against the Group B Streptococcus (GBS) Srr2 protein. The study aims to determine whether citric acid can bind the Srr2 adhesin and prevent its involvement in the pathogenesis of GBS. The investigation also evaluates the pharmacokinetics and safety profile of citric acid; thus, it is a possible replacement for antibiotics, and the ultimate goal is to maintain the neonatal microbiome and reduce the degree of illness prevalence. Although citric acid exhibits good affinity with the Srr2 adhesin, its off-target effects should also be considered. Citric acid is a natural metabolite, and although the models in silico may suggest that it interacts less with human proteins, experimental research is still required to reveal the binding pattern specificity. Any off-target effects, especially with other bacterial adhesins or human proteins, may affect the overall effectiveness and safety of citric acid, especially in the sensitive neonatal environment. These interactions should be tested through experimental studies to optimize the therapeutic approach and reduce the unwanted side effects.

This is the first study to assess postbiotic metabolites and particularly citric acid against the Srr2 virulence factor of Group B Streptococcus through integrated molecular docking, ADME/T profiling, and molecular dynamics simulation analysis.

2. MATERIALS AND METHODS

2.1. Study Design

An in silico antivirulence study was conducted targeting the Srr2 adhesin of Streptococcus agalactiae. The workflow included target preparation, domain and pocket mapping, ligand selection and preparation, blind docking with an AutoDock Vina–based server, interaction profiling, in silico ADME and toxicity prediction, and molecular dynamics simulations, followed by stability analyses using root mean square deviation, root mean square fluctuation, and radius of gyration. Software versions, operating system details, parameters, and scripts were recorded and are provided in the Data and Code Availability section. This design was selected to test whether postbiotic metabolites could bind an adhesin implicated in the pathogenesis of neonatal meningitis, thereby supporting an antivirulence approach that may lessen pressure on classical antibiotic resistance while remaining relevant to the clinical disease burden [11, 12].

2.2. Target Protein Selection

The serine-rich repeat protein Srr2 was selected because it mediates adhesion to host ligands, contributes to traversal of protective barriers, and supports biofilm formation, all of which are linked to invasive neonatal disease. The experimental Srr2 adhesion domain structure in the Protein Data Bank under PDB ID 4MBR, chain A, was used as the primary receptor for structure-based analyses. When required for completeness, missing regions were modelled and energy-minimised before downstream calculations. Selection of Srr2 as the target provided a mechanistic focus on a validated virulence determinant associated with meningitis development and poor outcomes in newborns [12].

2.3. Sequence Retrieval

The amino acid sequence corresponding to the Srr2 domain used in structural analyses was retrieved from curated databases and cross-checked against the NCBI resource to ensure consistency between the sequence record and the structural entry [13]. The FASTA file and the coordinate file were archived with the project materials to permit exact reproduction of all computational steps.

2.4. Physicochemical Properties and Three-Dimensional Structure

Physicochemical descriptors, such as molecular weight, theoretical isoelectric point, instability index, aliphatic index, and the grand average of hydropathicity are computed using the ExPASy ProtParam resource with default parameters and recorded access information [14]. As mentioned in the above discussion, the three-dimensional structure of the receptors was retrieved at the Protein Data Bank. Prior to docking and simulation, crystallographic ligands, and water molecules were eliminated and polar hydrogen, and a brief energy minimization to alleviate steric clashes were added. Where comparative modeling was necessary, publicly accessible modeling services like I-TASSER or AlphaFold were used to verify domain architecture and to provide information on how to handle unresolved regions [15]. All the prepared structures, intermediate files and logs were stored and deposited to facilitate verification.

2.5. Structural and Functional Domain Analysis

Protein visualization and inspection were performed in PyMOL. The receptor file was loaded, solvent and non-receptor entities were removed, and the final preparation was saved for uniform use across docking and simulation steps [16]. Functional domain annotation was carried out using InterPro, which identified an SDR-Ig domain spanning residues 47 to 145 and a fibrinogen-binding adhesion domain spanning residues 179 to 320 in the Srr2 construct used here. The contextualization of binding pockets and the interpretation of residue-level interactions were done using these annotations in docking and molecular dynamics analyses. Numbers were generated directly from PyMOL sessions, and domain boundaries were cross-tabulated with InterPro entries to ensure that the sequence features and the structural model are consistent [17].

2.6. Active Site Identification

Protein cavities and surface depressions were identified with CASTp, using the default solvent probe radius of 1.4 Å to compute pocket area and volume [18]. The pocket with the largest combined area and volume that overlapped the annotated adhesion domain was selected as the primary binding site for structure-based work. This decision maximized the number of druggable sub-pockets that would be found and was in agreement with the established domain activity. Pocket identifiers, centroids, surface meshes and the numerical descriptors were captured and reported with coordinates in Supplementary table S1 to allow precise recreation of the docking grid.

2.7. Ligand Selection and Preparation

Eight post-biotic-derived metabolites were determined by relevance as reported in the gastrointestinal and perinatal settings, safety indicators in general, and chain length and unsaturation. PubChem was also considered to identify any canonical structures and identifiers that were tracked and transformed into a downstream format [19]. The tautomers and protonation state of the ligands were standardized at physiological pH and the individual ligands were minimized at the gas phase in Chem3D version 12.0.2 and thus minimized conformational strain and enhanced stability of the pose during docking and simulation [20]. Last three-dimensional conformers and metadata were stored.

2.8. Drug-Likeness Screening and ADME/T Prediction

The Lipinski rule of five [21] thresholds of the molecular weight, lipophilicity, hydrogen-bond donors, and hydrogen-bond acceptors were used as drug-likeness filters along with Veber-style criteria of rotatable bonds and topological polar surface area to describe the constraints of oral exposure. Absorption distribution, metabolism, excretion and toxicity properties were then predicted on an established in silico platform. Examples of endpoints were intestinal absorption, binding to P-glycoprotein, volume of distribution, fraction unbound, blood-brain barrier indices, cytochrome P450 liabilities, total clearance, and the desired toxicities of hERG inhibition and hepatotoxicity, as suggested in methodological evaluations of the utility of ADMET prediction [22]. Outputs of full prediction and configuration files were left to enable independent verification.

2.9. Molecular Docking

The process of blind docking was carried out using a cavity-aware server that used the AutoDock Vina scoring functionality. The pocket centroids and surfaces obtained in CASTp were used in informing the construction of docking grids such that search boxes covered the adhesion-domain pocket chosen in Section 4.6. The a priori parameters were exhaustiveness, number of modes, energy window, and random seeds, which were logged in run logs. Each ligand was posed in multiple poses, and the best pose according to Vina score was kept, and ties were sorted by buried solvent-accessible surface area and size of hbond network, as it is customary to dock ligands [23, 24]. Input files (protein and ligand) were saved, grid parameters, seeds, and server outputs were saved, which allows executing the experiment precisely again.

2.10. Protein–Ligand Interaction Profiling

Interaction diagrams were prepared using LigPlot+, and hydrogen-bond cutoffs of 3.5 A were set to define how far the donor and acceptor were separated, and the hydrophobic contact used 120o as the angle cutoff [25]. The five most persistent polar and nonpolar contacts were tabulated in each top complex and interatomic distances were also stored and three-dimensional scenes were stored as PyMOL sessions. This reporting made comparisons between residue-levels across ligands feasible and provided the mechanistic interpretation of docking and molecular dynamics findings.

2.11. Lead Compound Identification

The prioritization of leads was based on a predetermined decision criteria that included docking rank and pose reproducibility, density of interactions in the adhesion domain and characteristics of ADME/T liabilities. Those that did not pass drug-likeness filters or those that raised significant safety issues were eliminated. According to structural and pharmacokinetic rationale, the resulting short list was put forward to simulation.

2.12. Reference Antibacterial Drug Selection

A professional body of drug information listed a clinically applicable antibacterial agent in the management of neonatal Group B Streptococcus that creates a best practice point of like-for-like comparisons [25]. The comparator was prepared to the same file preparation, docking protocol and ADME/T predictions as the postbiotic candidates, hence any differences were related to the properties of the ligands but not methodological alterations.

2.13. Molecular Dynamics Simulations

All complexes to be simulated were made in explicit water in the presence of neutralizing ions and physiological ionic strength. When the charge derivation was done, protein parameters were deduced to a modern biomolecular force field, and ligand parameters were deduced to a general small-molecule force field. The minimization of energy, gentle heating to 310 K, and isothermal isobaric equilibration were followed by a 100 ns production run. It used the particle mesh Ewald method to solve long range electrostatics; hydrogens bonds were constrained so that a two fs timestep could be used; temperature and pressure were stabilized with a Langevin thermostat and a suitable barostat. Separate replicate paths were created using different seeds. Protein backbone root mean square deviation, ligand heavy-atom root mean square deviation, per-residue root mean square fluctuation, radius of gyration and hydrogen-bond occupancy were analyzed. The input scripts, parameter files, seeds and analysis notebooks were deposited to enable full reproducibility. Three individual 100-ns molecular dynamics simulations were carried out on each protein-ligand complex with various random seeds. The trends in RMSD, RMSF and radius of gyration were qualitatively compared between replicates to determine convergence and reproducibility and trend patterns of stability were consistent within the lead complex.

3. RESULTS

3.1. FASTA Sequence Retrieval

Fasta sequence of the Srr2 protein with high concentration of the amino acid Serine was obtained using NCBI with the accession number 4MBR_A GI: 557129588. The choice of this protein is based on its pathogenicity and virulence-causing factor.

3.2. Physicochemical Characterization of Target Protein

Protparam was used to forecast the chemical and physical characteristics of the Srr2 protein. Further navigating a molecular weight of 38442.52 points. The target protein was acidic; its calculated PI was 4.82, less than 7. The extinction coefficient, with specific values of 46300 and 45930, shows the protein’s relatively high light absorption capacity, indicating that it is a more stable protein. The target protein’s instability index is 25.64; less than 40 indicates protein stability [26]. The target protein’s high aliphatic index of 78.14 indicates higher thermostability. GRAVY values around -0.522 suggest a relatively equal balance between hydrophobic and hydrophilic amino acids within the protein sequence. It indicates that Srr2 has a moderate level of both hydrophobic and hydrophilic characteristics. To some extent, moderate GRAVY values, such as -0.522, can interact with both aqueous and non-aqueous environments. This can benefit proteins with functions that involve interfaces between water and lipids within the cell [27].

3.3. 3D Structure Prediction of Target Protein

The 3D structure of serine-rich repeat protein 2 Srr2 was taken from the PDB under ID 4MBR. PyMOL helped to create the protein structure by eliminating any ligands and water molecules that might have been present (Fig. 1a). After removing the ligands and other atoms, the researchers added the missing polar hydrogens. To achieve a stable conformation by avoiding overlaps, the energy reduction for the structure was carried out, and the amended file was saved in PDB format [28].