0% READ
FULL TEXT
INTRODUCTION
Klebsiella pneumoniae is a Gram-negative pathogen responsible for a wide range of infections, including pneumonia, septicemia, and urinary tract infections, particularly in immunocompromised patients.1 The emergence of Multidrug-Resistant (MDR) strains, especially those producing Extended-Spectrum β-Lactamases (ESBLs) and carbapenemases, has become a significant global health concern.2,3 The HS11286 strain, isolated from a sputum sample in Shanghai, China in 2011, is a notable example of a Carbapenemase-Producing K. pneumoniae (CPE) strain with a high level of antibiotic resistance, making it a critical target for research.4–6 The rapid spread of MDR K. pneumoniae strains has severely limited treatment options, leading to increased morbidity, mortality, and healthcare costs.7,8 Despite advances in genomic research, a significant portion of the K. pneumoniae proteome, particularly Hypothetical Proteins (HPs) with unknown functions, remains uncharacterized. These HPs may play crucial roles in bacterial survival, pathogenesis, and antibiotic resistance, yet their potential as drug targets remains largely unexplored.9,10 This study aims to computationally annotate and experimentally validate the functions of HPs in K. pneumoniae HS11286 to identify novel drug targets. By leveraging bioinformatics tools, we seek to classify these proteins, predict their subcellular localization, and assess their potential as therapeutic targets. Additionally, we aim to validate the role of selected HPs in antibiotic resistance through experimental approaches.
Using a combination of sequence retrieval, functional prediction, subcellular localization analysis, subtractive genomics, and structural modeling, we systematically analyzed 974 HPs from the K. pneumoniae HS11286 genome. Experimental validation was performed using Polymyxin B and the PmrD protein to assess the role of selected HPs in antibiotic resistance. This integrated approach provides a comprehensive framework for identifying and validating potential drug targets in K. pneumoniae.
MATERIALS AND METHODS
Bacterial Strain and Culture Medium
The Klebsiella pneumoniae HS11286 strain was utilized in this study. The bacteria were cultured in Mueller-Hinton (MH) medium under standard laboratory conditions.
Chemicals and Reagents
Polymyxin B (Sigma-Aldrich) was served as a positive control drug for antibacterial activity assays. Other reagents included sterile distilled water, Phosphate-Buffered Saline (PBS), and Dimethyl Sulfoxide (DMSO) for drug dilution.
Experimental Materials
96-well plates (Corning) for bacterial growth and drug susceptibility assays. Micropipettes (Eppendorf) for precise liquid handling. Incubator (Thermo Scientific) set at 37ºC for bacterial culture. Microplate reader (BioTek) for measuring Optical Density (OD) at 600 nm to assess bacterial growth.
Methods
The list of tools used throughout this functional annotation of K. pneumoniae shows in Table 1. These tools are available in online as open source where it can use anytime without the restriction of time and cost. In the table above, each website link to the tools is provided together with the name of the tools. Workflow of computational pipeline for annotating and finding drug targets among HPs was showed in Figure 1.
Sequence Retrieval and Preprocessing
The complete proteome of Klebsiella pneumoniae HS11286 was retrieved from the National Center for Biotechnology Information (NCBI). A total of 1,694 Hypothetical Proteins (HPs) were identified, and their sequences were obtained in FASTA format from UniProt. Redundant sequences were removed using the Cluster Database at High Identity with Tolerance (CD-HIT) tool, resulting in 974 non-redundant HPs for further analysis.
Functional Prediction
Functional domains of the 974 HPs were predicted using multiple bioinformatics tools: CDART: Identified conserved domains through evolutionary distance analysis. InterProScan: Classified protein families and predicted functional sites. SVM-Prot: Predicted protein domains using a support vector machine-based approach. Pfam: Provided comprehensive domain and family classification.
Subcellular Localization Prediction
Subcellular localization of the HPs was predicted using: CELLO: A two-level support vector machine tool for bacterial and eukaryotic localization. PSLpred: Specifically designed for Gram-negative bacteria. TMHMM and HMMTOP: Predicted transmembrane helices and membrane protein propensity. SignalP: Identified signal peptides involved in non-classical secretory pathways.
Subtractive Genomics
To identify essential and non-toxic HPs, we performed a stepwise filtering process: Non-homology to human proteins: The 974 HPs were compared against the human proteome (RefSeq GRCh38) using BLASTp (E-value cutoff: 10⁻⁴), excluding any HP with >30% sequence identity to human proteins to ensure minimal off-target effects.
Essentiality screening: Remaining HPs were analyzed against the Database of Essential Genes (DEG) using an E-value threshold of 10⁻¹⁰⁰ and coverage >50% to confirm bacterial essentiality. Virulence and druggability: Essential HPs were further filtered by: VirulentPred score >0.7 (indicating high virulence potential), DrugBank druggability classification (presence of known drug-binding domains). This pipeline ensured the selection of targets that are essential for bacterial survival, non-homologous to humans, and pharmacologically tractable.
Virulence Prediction
The virulence potential of the HPs was assessed using the VirulentPred tool, which classified proteins as virulent or non-virulent based on their sequence features.
Structural Modeling
Three-dimensional structures of the selected HPs were modeled using the Phyre2 server, which employs homology detection and advanced modeling techniques to predict protein structures and ligand-binding sites. Model quality was assessed by: Ramachandran plot analysis (via PROCHECK) to evaluate backbone torsion angles; QMEAN score (Z-score >-4.0 considered acceptable); Template identity threshold (>30% sequence identity to known structures). Models with >90% residues in favored regions (Ramachandran) and QMEAN Z-score > -3.0 were retained for further analysis.
Drug Target Prediction
The druggability of the HPs was evaluated using the DrugBank database, which contains over 8,000 drug entries, including small molecules and biologics. Proteins with known drug interactions were classified as putative drug targets.
Experimental Validation
Bacterial Culture: Klebsiella pneumoniae HS11286 was cultured in Mueller-Hinton (MH) medium at 37ºC until reaching the logarithmic growth phase.
Concentration Selection: Polymyxin B concentrations (0-100 μg/ mL) were chosen based on:
Clinical breakpoints for K. pneumoniae (EUCAST guidelines). Preliminary dose-response experiments showing 50% inhibition at ~25 μg/mL. PmrD protein concentration (100 μg/mL) was optimized to: Match physiological levels in resistant strains. Avoid non-specific growth effects (validated in control experiments)
Antibacterial Activity Assay: Bacterial suspensions were adjusted to an appropriate concentration and inoculated into 96-well plates. Polymyxin B and PmrD protein were added at varying concentrations (0-100 μg/mL), a control group without drugs was included and bacterial growth was monitored using a microplate reader.
Control Experiments: To rule out non-specific effects, the following controls were included: Negative control: Bacteria cultured without Polymyxin B or PmrD protein. Solvent control: Bacteria treated with DMSO (the solvent for Polymyxin B) at the highest concentration used. Protein control: Bacteria treated with PmrD protein (100 μg/mL) alone to exclude growth effects from the protein itself. Positive control: Bacteria treated with Polymyxin B alone (without PmrD).
Growth Measurement: Plates were incubated at 37ºC for 24 hr, and bacterial growth was monitored by measuring OD at 600 nm using a microplate reader, and the bacterial growth inhibition rate was calculated for each treatment group.
Statistical Analysis
Bacterial growth inhibition rates were analyzed using: Two-way ANOVA with Tukey's post-hoc test to compare means across Polymyxin B concentrations (0-100 μg/mL) and treatment groups (Control vs. PmrD-treated). Shapiro-Wilk test (p>0.05) confirmed normal distribution of residuals. Equal variance assumption was verified via Levene's test. Data are presented as mean±SD (n=3 biological replicates). All analyses were performed in GraphPad Prism 9.0.
| Sl. No. | Tools | URL |
|---|---|---|
| 1. | NCBI | |
| 2. | Uniprot | |
| 3. | CD-HIT | |
| 4. | CDART | |
| 5. | Interproscan | |
| 6. | SVMprot | |
| 7. | Pfam | |
| 8. | PSLpred | |
| 9. | CELLO | |
| 10. | TMHMM | |
| 11. | SignalPeptide | |
| 12. | HMMTOP | |
| 13. | blastp | |
| 14. | DEG | |
| 15. | VirulentPred | bioinfo.icgeb.res.in/virulent |
| 16. | Phyre2 | |
| 17. | DrugBank |
RESULTS AND ANALYSIS
Functional Classification of Hypothetical Proteins (HPs)
Among the 974 HPs analyzed, 732 were functionally classified using CDART, InterProScan, SVM-Prot, and Pfam tools. The functional categories are summarized in Figure 2. The majority of HPs were classified as enzymes (239 HPs), followed by proteins with unknown functions or domains of unknown function (DUF) (204 HPs). Other significant categories included binding proteins (68 HPs), transporters (41 HPs), and bacteriophage-related proteins (14 HPs). Notably, 242 HPs could not be functionally predicted, highlighting the need for further experimental validation.
Subcellular Localization of HPs
Subcellular localization analysis using CELLO revealed that the majority of HPs (528 out of 974) were predicted to localize in the cytoplasm (Figure 3). Other localization sites included the inner membrane (196 HPs), periplasmic space (161 HPs), outer membrane (53 HPs), and extracellular space (36 HPs). The cytoplasmic localization of 528 HPs suggests their potential as drug targets, as cytoplasmic proteins are often involved in essential metabolic and regulatory pathways.
Subtractive Genomics and Essential Protein Identification
Subtractive genomics analysis identified 434 HPs with no homology to the human proteome, ensuring non-toxicity to humans (Table 2). Further screening against the Database of Essential Genes (DEG) revealed 101 HPs that are essential for bacterial survival but non-toxic to humans. These proteins represent potential drug targets for K. pneumoniae.
Virulence Prediction
using VirulentPred classified 30 out of 101 HPs as virulent (Figure 4). These virulent HPs are likely involved in pathogenicity and may serve as promising targets for therapeutic intervention. The remaining 71 HPs were classified as non-virulent, suggesting roles in non-pathogenic cellular
Sl. No. Tools URL processes. HS11286 Classification of HPs as virulent (30 proteins) or non-virulent (71 proteins) using the VirulentPred tool.
Structural Modeling of Virulent HPs
Three-dimensional structures of the 30 virulent HPs were modeled using Phyre2 (Figure 5). The models revealed diverse structural features, including helix-turn-helix domains, Polymyxin Resistance Protein (PmrD) structures, and DNA-binding domains. These structural insights provide a foundation for understanding the functional mechanisms of these proteins and their potential as drug targets.
Model Validation Metrics
All 30 predicted structures passed quality checks:Ramachandran statistics: 92.5±3.2% residues in favored regions (range: 88-97%); QMEAN Z-scores: -2.1±0.8 (mean±SD), indicating models were within the expected range of native structures; Template coverage: 67±12% of residues modeled at >30% sequence identity. The PmrD model (UniProt ID: A0A0H3GGX7) showed particularly high quality (96% favored residues, QMEAN=-1.8).
Druggability Assessment
Druggability analysis using the DrugBank database identified all 30 virulent HPs as putative drug targets (Table 3). Each HP was associated with known drug interactions, suggesting that these proteins could be targeted by existing or novel therapeutics. For example, the PmrD protein (UniProt ID: A0A0H3GGX7) was identified as a key resistance factor, with potential interactions with Polymyxin B and other antibiotics.
Experimental Validation of PmrD-Mediated Polymyxin B Resistance
To validate the role of PmrD in antibiotic resistance, we compared the growth inhibition rates of K. pneumoniae HS11286 treated with Polymyxin B in the presence and absence of the PmrD protein. Statistical analysis (two-way ANOVA with Tukey's test) revealed that PmrD significantly increased bacterial resistance to Polymyxin B at all tested concentrations (p<0.01). As shown in Table 4, the PmrD-treated group exhibited significantly higher resistance to Polymyxin B compared to the control group across all tested concentrations (25-100 µg/mL). For instance, at 50 µg/mL Polymyxin B, the inhibition rate was 58.7±5.2% in the PmrD-treated group versus 39.6±4.1% in the control (p<0.001).
Control Experiments Results
All control groups showed expected growth patterns: Negative and solvent controls exhibited normal growth (OD600=1.2±0.1 after 24 hr). PmrD protein alone did not affect growth (OD600=1.1±0.2 vs negative control, p=0.45). These results confirm that the observed inhibition was specifically due to Polymyxin B and its interaction with PmrD.
Data analyzed by two-way ANOVA (factors: ⟦R1|1⟧ Polymyxin B concentration, ⟦R2|2⟧ Treatment) with Tukey's multiple comparisons. p-values indicate significant differences between Control and PmrD-treated groups at each concentration. *Controls: Negative (no treatment), solvent (DMSO), and PmrD-only controls showed <5% growth inhibition in all experiments.
Molecular Mechanism of PmrD-Mediated Polymyxin B Resistance
The experimental results demonstrate that PmrD significantly enhances K. pneumoniae resistance to Polymyxin B (Table 4). At the molecular level, PmrD is known to regulate the pmrAB two-component system, which controls modifications to the bacterial outer membrane. Specifically, PmrD stabilizes the phosphorylated state of PmrA (response regulator), leading to upregulation of genes such as arnT and pmrE. These genes mediate the addition of 4-amino-4-deoxy-L-arabinose (L-Ara4N) to lipid A, reducing the negative charge of the outer membrane and thereby decreasing Polymyxin B binding. Our structural modeling of PmrD (Figure 5) further supports its interaction with PmrA, suggesting a direct role in this regulatory cascade.
| Description | No. of proteins |
|---|---|
| Total no. of hypothetical proteins | 1694 |
| Total proteins above 100 sequences length/cd-hit paralogue removal | 974 |
| Sequence predicted with cytoplasmic | 528 |
| Human non-homologous proteins (blastp) (E-value 10-4) | 434 |
| Essential proteins found in DEG (E- value 10-100) | 101 |
| Sl. No. | Uniprot id | Domain function | Drug Bank ID | Drug Bank (Novel/ Putative) |
|---|---|---|---|---|
| 1. | A0A0H3GGX2 | Helix-turn-helix Domain (Binding Protein) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 2. | A0A0H3GGX7 | Polymyxin resistance protein PmrD (Factor, regulator and hormones) | DB01413 DB01208 DB01211 DB01155 DB00207 | Putative |
| 3. | A0A0H3GP51 | YbaB/EbfC DNA-binding (Binding protein) | DB01413 DB00760 DB01208 DB01211 DB00207 | Putative |
| 4. | A0A0H3GKQ2 | NinB protein (Bacteriophage related protein) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 5. | A0A0H3GL86 | Thioesterase superfamily protein (Enzyme) | DB01413 DB01208 DB01155 DB00303 DB00567 | Putative |
| 6. | A0A0H3GKX5 | DUF2813 (Unknown function/DUF) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 7. | A0A0H3GQT0 | Macrodomain Ter protein, MatP (Cellular process regulator protein) | DB01413 DB01208 DB01211 DB00760 DB00207 | Putative |
| 8. | A0A0H3GLE0 | MsyB protein (Miscellaneour function) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 9. | A0A0H3GLM7 | GAF domain-like (Binding protein) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 10. | A0A0H3GRP9 | None predicted | DB01155 DB01413 DB09060 DB01208 DB01211 | Putative |
| 11. | A0A0H3GRR4 | DUF697 (Unknown function/DUF) | DB01413 DB01208 DB01211 DB00760 DB00207 | Putative |
| 12. | A0A0H3GSR9 | DUF1398 (Unknown function/ DUF) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 13. | A0A0H3GSD7 | Smr domain (Cell cycle related protein) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 14. | A0A0H3GT39 | BLUF domain (Binding protein) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 15. | A0A0H3GTD8 | PAS domain (Signalling protein) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 16. | A0A0H3GXV7 | PAS domain (Signalling protein) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 17. | A0A0H3GUW6 | Thioesterase Superfamily (Enzyme) | DB01413 DB01208 DB01155 DB09060 DB00303 | Putative |
| 18. | A0A0H3H026 | DUF535 (Unknown function/ DUF) | DB01208 DB09060 DB01155 DB01413 DB00467 | Putative |
| 19. | A0A0H3GVT2 | Glycosyl transferase (Enzyme) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 20. | A0A0H3GTG7 | Arsenate reductase-like (Enzyme) | DB01413 DB01208 DB01211 DB00760 DB00207 | Putative |
| 21. | A0A0H3GTW8 | Periplasmic sensor domain (Signalling protein) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 22. | A0A0H3GS84 | Uncharacterised protein family YgbA (Unknown function) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 23. | A0A0H3GSD2 | tRNA pseudouridine synthase C (Enzyme) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 24. | A0A0H3GXJ0 | Pilus retraction protein PilT (Binding protein) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 25. | A0A0H3GXR7 | DUF1249 (Unknown function/DUF) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 26. | A0A0H3GY16 | RNA-binding protein YhbY (Binding protein) | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| 27. | A0A0H3GY32 | P-loop ATPase protein family (Enzyme) | DB01413 DB01208 DB01211 DB00760 DB00207 | Putative |
| 28. | A0A0H3GWN4 | TusC/DsrF (Transport protein) | DB01413 DB01208 DB01155 DB00303 DB00567 | Putative |
| 29. | A0A0H3GRH9 | HTH-like domain (Binding protein) | DB01413 DB09060 DB01208 DB01155 DB00303 | Putative |
| 30. | A0A0H3GZF9 | Resolvase, N terminal domain Enzyme | DB01413 DB09060 DB01208 DB01211 DB00207 | Putative |
| Polymyxin B Concentration (µg/mL) | Control Group Inhibition Rate (%) | PmrD-Treated Group Inhibition Rate (%) | Standard Deviation (Control) | Standard Deviation (PmrD) | |
|---|---|---|---|---|---|
| 0 | 0.0±0.0 | 0.0±0.0 | 0.0 | 0.0 | - |
| 25 | 18.2±2.8 | 33.5±3.6 | 2.8 | 3.6 | 0.003 |
| 50 | 39.6±4.1 | 58.7±5.2 | 4.1 | 5.2 | <0.001 |
| 75 | 54.3±6.0 | 73.8±7.3 | 6.0 | 7.3 | 0.001 |
| 100 | 68.9±7.5 | 84.2±6.8 | 7.5 | 6.8 | <0.001 |
DISCUSSION
This study provides a comprehensive functional annotation of Hypothetical Proteins (HPs) in Klebsiella pneumoniae HS11286 and identifies 30 potential drug targets through integrated computational and experimental approaches. The functional classification of 974 HPs revealed a diverse range of proteins, with enzymes and binding proteins being the most abundant categories (Figure 2). This finding aligns with previous studies highlighting the importance of enzymatic and binding activities in bacterial survival and pathogenesis.11,12 The predominance of cytoplasmic localization among the HPs (Figure 3) further supports their potential as drug targets, as cytoplasmic proteins often play critical roles in metabolic and regulatory pathways.13 Subtractive genomics analysis identified 101 HPs that are essential for bacterial survival but non-toxic to humans, providing a robust foundation for drug target discovery. Among these, 30 HPs were classified as virulent using VirulentPred (Figure 4), suggesting their involvement in pathogenicity. Structural modeling of these virulent HPs using Phyre2 (Figure 5) revealed diverse structural features, including helix-turn-helix domains and Polymyxin Resistance Protein (PmrD) structures. The high validation scores (mean QMEAN Z-score=-2.1, >92% favored Ramachandran regions) support the reliability of our models for docking studies, consistent with recent benchmarks showing Phyre2's accuracy for bacterial proteins.
These findings are consistent with previous reports on the role of HPs in antibiotic resistance and virulence.14, 15 Experimental validation using Polymyxin B and the PmrD protein demonstrated that K. pneumoniae HS11286 exhibits significant resistance to Polymyxin B, a resistance further enhanced by the presence of PmrD (Table4). The inclusion of multiple controls (negative, solvent, and protein-only) confirmed that the observed resistance phenotype was specifically mediated by PmrD's interaction with Polymyxin B, rather than non-specific effects. This suggests that PmrD may modulate cell membrane permeability or interfere with drug binding, contributing to antibiotic resistance. These results are in line with studies on the role of resistance proteins in Gram-negative bacteria.16,17 Our identification of 30 druggable Hypothetical Proteins (HPs), including PmrD, opens new avenues for combating MDR K. pneumoniae. Specifically:Targeting PmrD-mediated resistance: Inhibiting PmrD could restore Polymyxin B sensitivity in resistant strains by blocking lipid A modifications. This approach may extend the clinical utility of last-resort antibiotics. Exploiting essential non-human homologs: The 101 essential HPs with no human homology (Table 2) represent high-priority targets for narrow-spectrum drug development, minimizing off-target toxicity. Multi-target strategies: Simultaneously targeting virulent HPs (e.g., PmrD) and resistance-related enzymes (e.g., thioesterases in Table 3) may prevent compensatory mutations. These strategies could complement existing therapies, particularly against carbapenemase-producing strains where treatment options are severely limited. Future studies should focus on molecular docking and drug design to develop inhibitors targeting these proteins, particularly those involved in antibiotic resistance mechanisms.
CONCLUSION
In conclusion, our study not only provides a comprehensive functional annotation of 974 hypothetical proteins in K. pneumoniae HS11286 but also identifies 30 high-confidence drug targets, including the resistance regulator PmrD. Crucially, we demonstrate that PmrD enhances Polymyxin B resistance by stabilizing the PmrAB system, leading to lipid A modifications-a mechanism that can be pharmacologically disrupted to restore antibiotic susceptibility. Among the prioritized targets, the 101 essential HPs with no human homology offer exceptional potential for developing narrow-spectrum therapies with minimal off-target effects. Moreover, targeting multiple virulence-associated HPs (e.g., PmrD and thioesterases) may prevent compensatory resistance evolution, a critical advantage against rapidly adapting MDR strains. These findings establish a rational framework for designing next-generation antimicrobials, particularly against carbapenem-resistant K. pneumoniae where treatment options are exhausted. Future work should advance lead compound discovery through structure-based drug design and in vivo validation of these targets.
SUMMARY
This study investigates the potential of Hypothetical Proteins (HPs) in Klebsiella pneumoniae HS11286 as novel drug targets. Using bioinformatics tools, we analyzed 974 HPs for their functions, subcellular localization, and drug target potential. Techniques such as subtractive genomics and virulence prediction were employed to identify essential and non-toxic proteins. Structural modeling and druggability assessments were conducted on selected HPs. Experimental validation using Polymyxin B and the PmrD protein demonstrated that PmrD significantly enhances K. pneumoniae's resistance to Polymyxin B, highlighting its role in antibiotic resistance mechanisms. The study provides a comprehensive functional annotation of HPs and identifies 30 potential drug targets, paving the way for the development of new therapies against multidrug-resistant K. pneumoniae. Future research will focus on molecular docking and high-throughput screening to validate and optimize these targets for clinical applications.
