Immunoinformatics-driven design and computational analysis of a multiepitope vaccine targeting uropathogenic Escherichia coli

Abstract

Urinary tract infections (UTIs), largely caused by uropathogenic Escherichia coli (UPEC), are increasingly resistant to antibiotics and frequently recur. Using immunoinformatics, we designed a multiepitope peptide vaccine targeting UPEC virulence factors, including iron acquisition systems and adhesins. The construct features 12 cytotoxic T lymphocyte epitopes, six helper T lymphocyte epitopes, and six B-cell epitopes,and isoptimized for high antigenicity, immunogenicity, nontoxic, and low allergenic potential. Molecular docking and 0.4-µs molecular dynamics simulations revealed the molecular mechanism of theinteraction of the vaccine with Toll-like receptor 4 and a favorable binding energy of − 41.83 kcal/mol using an implicit solvation model. These promising in silico results suggest the potential efficacy of the vaccine in preventing UPEC infections and underscore immunoinformatics as a powerful tool for addressing antibiotic-resistant UTI pathogens.

Graphical Abstract

Supplementary information

The online version contains supplementary material available at 10.1007/s40203-024-00288-z.

Introduction

Urinary tract infections (UTIs) are the most common bacterial infections seen by healthcare practitioners, affecting more than 150 million people globally each year. Although several pathogens can cause UTIs, UPEC (uropathogenic E. coli) is responsible for the majority of urinary tract infections (Flores-Mireles et al. 2015; Ulett et al. 2013). UTIs result from microbial invasion of the epithelial tissue of the urinary tract, triggering an inflammatory response (Foxman 2010).

UPEC virulence factors include iron acquisition systems and adhesins, such as type 1 pili and FimH. These virulence factors are responsible for invasion, colonization, and protection from antibiotics and host defenses (Kisiela et al. 2015). FimH, with its pilin domain at the C-terminus and lectin domain at the N-terminus, facilitates UPEC binding to host cell receptors (Mossman et al. 2008). Several studies have shown that iron acquisition genes (Bakhtiari et al. 2016; Garcia et al. 2011) and the siderophore receptor IutA are highly upregulated in UPEC strains during UTIs. IutA is more prevalent among pathogenic serotypes and plays a critical role in iron binding (Bakhtiari et al. 2016; Hagan and Mobley 2007).

While antimicrobial medications are the usual treatment for UTIs, the emergence of antibiotic-resistant pathogens has limited their effectiveness. The high cost of pharmaceuticals and the increasing incidence of UTIs highlight the need for vaccines (Akram et al. 2007; Bien et al. 2012; Moriel et al. 2010). In the 21st century, drug discovery has not kept pace with the rapid rise in antibiotic resistance, making bacterial infections difficult to manage (Medina and Pieper 2016). In regions with limited resources, access to healthcare is restricted, and the high cost of new-generation antibiotics further complicates treatment options (Akram et al. 2007).

Individuals with a history of UTI are at high risk for recurrence, making therapeutic intervention difficult (Russell et al. 2023). Understanding UPEC resistance mechanisms is crucial for guiding prescription decisions for healthcare providers (Kaur et al. 2014). Although antibiotics reduce symptoms, UPEC can persist, leading to relapses that are difficult to manage (Blango and Mulvey 2010).

Immunoinformatic approaches have advanced the development of multiepitope vaccines, which contain both B-cell and T-cell epitopes to induce broad immune responses with effective translational applications (Dey et al. 2023; Paul et al. 2023). The integration of adjuvants can improve the immunogenicity and durability of immune responses (Zhang 2018; Umitaibatin et al. 2023; Sarvmeili et al. 2024). In this study, we designed a multiepitope vaccine targeting UPEC virulence factors, focusing mainly on iron acquisition systems, with corresponding adhesins and toxins. The multiepitope vaccine construct interacts with Toll-like receptor 4 (TLR4) because TLR4 plays a crucial role in the immune system by recognizing pathogens and initiating immune responses. In the context of vaccine design, specifically multiepitope constructs, TLR4 activation is desirable because it triggers the innate immune response, which is essential for developing a strong adaptive immune response. We used computational methods to predict B cell, MHC class I, and MHC class II epitopes from UPEC antigens including IutA, FimH, IreA, IroN, FyuA and ChuA. The designed vaccine candidate was docked with the TLR4 receptor, and molecular dynamics simulations were performed to assess binding affinity.

Methodology

Protein sequence retrieval

The protein sequences of UPEC virulence factors with accession numbers were retrieved from NCBI (www.ncbi.nlm.nih.gov): IutA (AAN82071.1), FimH (AAN83822.1), IreA (WP_000792545.1), IroN (AAN79707.1), FyuA (AAP70282.1), and ChuA (AAG44838.1).

Cytotoxic T-lymphocytes (CTL) epitope mapping

The CTL epitopes were predicted via the NetCTL 1.2 server (http://www.cbs.dtu.dk/services/NetCTL/) (Larsen et al. 2007), which uses an artificial neural network algorithm. Major histocompatibility complex-1 (MHC-I)-binding epitopes were predicted using a threshold value of 0.75, representing 0.97 specificity and 0.80 sensitivity. The predicted epitopes were then screened for toxicity via the ToxinPred server (http://crdd.osdd.net/raghava/toxinpred/multi_submit.php), followed by an assessment of immunogenicity via the Immune Epitope Database (IEDB) (http://tools.iedb.org/immunogenicity/) (Moutaftsi et al. 2006).

Helper T-lymphocyte (HTL) epitope mapping and population coverage

HTL epitopes were predicted via the MHC-II binding tool from the IEDB (http://tools.iedb.org/mhcii/) (Jensen et al. 2018). A six-allele HLA reference set was selected, including and HLA-DRB101:01, HLA-DRB302:02, HLA-DRB501:01, HLA-DRB301:01, HLA-DRB302:02, and HLA-DRB115:01. The epitopes were then screened for toxicity, antigenicity, and allergenicity via ToxinPred (Sharma et al. 2022), VaxiJen v.2.0 (Doytchinova and Flower 2007), and AllerTOP v.2 (Dimitrov et al. 2014). Epitopes that overlapped were removed.

Furthermore, CTL predictions targeted the HLA allele reference set, covering more than 97% of the global population (Weiskopf et al. 2013). HTL epitopes recognize HLA alleles with population coverage exceeding 99% (Greenbaum et al. 2011). HTL epitopes were also screened via the IFNepitope server to predict their ability to induce interferon-gamma (INF-γ) via Motif and SVM hybrid method (Dhanda et al. 2013).

B-cell epitope mapping and conservancy prediction

Linear B-cell epitopes were predicted via the ABCpred server (https://webs.iiitd.edu.in/raghava/abcpred/index.html) (Saha and Raghava 2006), which employs a recurrent neural network (RNN) technique. The epitopes were selected on the basis of a 0.51 threshold value (Saha and Raghava 2006). The VaxiJen 2.0 (https://www.ddg-pharmfac.net/vaxijen/VaxiJen/VaxiJen.html) ToxinPred, and AllerTOP v.2 (https://www.ddg-pharmfac.net/AllerTOP/index.html) servers were used to further assess antigenicity, toxicity, and allergenicity, respectively. This was followed by scrutinizing these epitopes for conservancy among the homologous strains viatools in the IEDB (http://tools.iedb.org/conservancy/ ).

Multiepitope vaccine integration and processing

The identified epitopes were linked via AAY and GPGPG linkers. The CTL epitopes were joined via AAY linkers, whereas the HTL and B-cell epitopes were connected via GPGPG linkers. The PADRE sequence and β-defensin adjuvant were linked via EAAAK linkers. The addition of linkers facilitates the effective separation of epitopes (Sethi et al. 2024), and increases antigen processing, presentation, and vaccine efficacy. The construct’s antigenicity, allergenicity, toxicity, and immunogenicity were evaluated via ProtParam (Gasteiger et al. 2005).

Modeling and structure validation

The secondary structure of the vaccine construct was analyzed via the PROTEUS2 server (Montgomerie et al. 2008). A 3D structure was generated via I-TASSER (Zhou et al. 2022) and refined via GalaxyRefine (Heo et al. 2013). Quality was validated via ProSA-web (Wiederstein and Sippl 2007) and the SAVES server.

Molecular docking and visualization

The ClusPro 2.0 server (https://cluspro.org/login.php) (Kozakov et al. 2017) was used to dock the vaccine construct with the TLR4 receptor (PDB ID: 4G8A). Molecular docking studies were conducted using ClusPro 2.0 to investigate the interaction between the multiepitope vaccine construct and TLR4. ClusPro’s default parameters were employed, utilizing an FFT-based algorithm to explore protein-protein orientations by scoring van der Waals and electrostatic interactions. A total of 30,000 potential docking models were generated, with clustering applied to the top 1,000 models, resulting in ten representative clusters based on the most populated configurations. The model with the lowest energy was selected for further analysis, providing insights into the structural compatibility and potential interactions essential for assessing the immunogenic potential of the vaccine candidate.

Molecular dynamics

The screened hit complexes were subjected to molecular dynamics (MD) simulations via the AMBER 22 package. The FF99SB force fielde was used for the protein-protein complex, respectively, as described by Case et al. (2005; Marques et al. 2021). For solvation, TIP3P water models were used, and the systems were neutralized by adding Na+ ions via the tLEap input script from the AmberTools 23 package. The Ewald method with a particle mesh was employed to represent long-range electrostatic interactions (Essmann et al. 1995). The SHAKE algorithm (Miyamoto and Kollman 1992) was used to constrain the length of covalent bonds.. A Langevin thermostat was used to maintain the system temperature at 310 K, and a 2.0-fs time step was used in all MD configurations.

A 100,000-step minimization and a 1-ns equilibration (NVT and NPT ensembles) were performed, followed by unconstrained 0.4 µs classical MD simulations using an NPT ensemble for the analyzed protein-protein complex. Furthermore, the MD protocol was combined with the generalized Born (MM-GBSA) molecular mechanics technique, which is reinforced by the hydrophobic solvent-accessible surface area term, as described in the works of Shityakov et al. (2017) and Dutta et al. (2021).

In silico cloning

The codon-optimized DNA sequence of the multiepitope construct was generated via the ExpOptimizer web server and tailored for expression in E. coli (strain K12) (Grote et al. 2005). The codon adaptation index (CAI) and GC content were evaluated, and the optimized sequence was subsequently cloned and inserted into the pET28a(+) vector via SnapGene software.

Protein-protein interface and stability analyses

To calculate the protein-protein interaction interface between the multiepitope vaccine construct and the receptor (TLR4), the PyMol Interface Calculator (http://ouchidekaiseki.com/en/interface.php) was utilized. The 3D structures of both the vaccine and receptor were obtained from previous molecular docking studies. Prior to interface calculation, the structures were prepared by adding any missing hydrogen atoms via Chimera software, followed by energy minimization via the AMBER99SB force field to ensure proper structural relaxation. The prepared structures were then uploaded to the PyMol Interface Calculator, which computes key parameters of the protein-protein interface, including the interface area, number of hydrogen bonds, salt bridges, and noncovalent interactions, offering insights into the strength and stability of the complex. To assess the stability of the multiepitope vaccine construct bound to TLR4, the standard Rosetta protocol was implemented, which involves calculating the energy in Rosetta energy units (REUs). This method predicts stability by correlating root-mean-square deviation (RMSD) values with Rosetta energy parameters, providing an estimation of the conformational stability of construct-TLR4.

HLA allele frequency analysis

The HLA allele frequency data were obtained from the HLA allele frequency search database, and the allele search interface was used to conduct a broad, inclusive analysis. Population filters were set to encompass all populations, countries, and regions, providing a globally representative dataset. To ensure comprehensive data collection, all available sources were included in the dataset selection. Resolution was set to capture all levels of genotyping specificity, accommodating data with varying detail. The frequency display settings were configured to include both positive and negative frequencies, allowing insight into both the presence and absence of each allele across diverse populations. Finally, the results were organized by allele in descending order of frequency to facilitate the identification of populations with the highest allele prevalence.

Results

Identification of cytotoxic T-cell and helper T-lymphocyte epitopes

The CTL epitopes were identified for six selected proteins using specified threshold values. From these, twelve epitopes were chosen on the basis of their high antigenicity, absence of allergenicity and toxicity,, and strong MHC-I binding affinity (Table 1). For MHC-II binding, six high-affinity HTL epitopes were predicted using the IEDB MHC-II server, and subsequently selected judging their antigenicity and the absence of allergenicity and toxicity (Table 2). The selected HTL epitopes for the human alleles HLA-DRB101:01, HLA-DRB302:02, HLA-DRB501:01, HLA-DRB301:01, HLA-DRB302:02, and HLA-DRB115:01 were located at positions 25–39, 41–55, 304–318, 442–456, 558–572, and 422–436, respectively. Interferon-gamma (INF-γ) plays a crucial role in the immune response by stimulating NK and cytotoxic T cells and helping to clear intracellular pathogens. Six HTL epitopes with positive INF-γ prediction scores were selected for vaccine construction.

Table 1.

List of the screened CTL epitopes that complement all the conditions of antigenicity, nontoxicity, and the absence of allergenicity and can bind effectively to supertype alleles, MHC-I A1

Protein	Peptide sequence	MHC binding affinity	Rescale binding affinity	C-terminal cleavage affinity	Transport efficiency	Prediction score
iutA	SSQQDTDQY	0.39	1.69	0.75	3.04	1.96
	LAAFLQSSY	0.27	1.16	0.95	3.14	1.46
FimH	CAKSQNLGY	0.25	1.06	0.92	2.88	1.34
	VVDLSTQIF	0.18	0.78	0.93	2.61	1.05
ireA	ELATQIPVY	0.17	0.71	0.97	2.8	1.0
	VTDRKSEDI	0.19	0.81	0.21	0.54	0.87
IroN	RSSEGANTY	0.56	2.37	0.98	3.1	2.68
	FSASFNWTL	0.4	1.71	0.97	1.07	1.9
FyuA	TTDDWVFNL	0.44	1.88	0.96	0.68	2.05
ChuA	TSATDLLRY	0.68	2.89	0.59	3.12	3.15
	TTDLFSLDV	0.62	2.61	0.93	0.04	2.75
	FAAATTMSY	0.46	1.95	0.71	2.76	2.19

Table 2.

The final screened HTL epitopes complement all the conditions of antigenicity, the absence of allergenicity and toxicity capable of inducing the IFN-γ immune response using SVM method

Protein	Allele (HLA)	Peptide sequence	Start	End	Percentile score	IFN-γ score
iutA	DRB1*01:01	GITMMRKKYMPRALG	25	39	0.01	0.2
FimH	DRB3*02:02	SANVYVNLAPAVNVG	41	55	0.38	0.3
ireA	DRB5*01:01	VYQEKVIREVKSGKK	304	318	0.63	0.4
IroN	DRB3*01:01	GLRFDYLSDSGGNFS	442	456	0.25	0.02
FyuA	DRB3*02:02	SELYHGNRVPFVPRY	558	572	0.5	0.5
ChuA	DRB1*15:01	NWLMLFGSYAQAFRA	422	436	0.8	0.14

Screening of linear B-cell epitopes and conservancy

The full-length sequences of IutA, FimH, IreA, IroN, FyuA, and ChuA were analyzed to predict B-cell epitopes. The ABCpred server identified the epitopes with cutoff values > 0.9, and six epitopes were selected for vaccine construction, all of which were non-allergenic, non-toxic, and highly antigenic (Table 3). The conservancy analysis predicted 100% conservancy of all the selected epitopes among the homologous strains.

Table 3.

The selected B-cell epitopes for the final construct with a binding score greater than 0.9

Protein	Peptide sequence	Start position	Predicted score
iutA	YSLSNKSVERNKDLTI	586	0.95
FimH	PFPTTSETPRVVYNS	143	0.92
ireA	GGIIQEHGKFGNSTTN	220	0.97
IroN	ALYIEDNIEPVPGTNI	463	0.96
FyuA	STVVSAPELSDAGVTA	71	0.97
ChuA	HISSSYSKQPGYGVND	638	0.91

HLA allele frequency

To enhance the study’s assessment of vaccine effectiveness across populations, a more detailed analysis of population coverage based on the HLA allele frequencies was performed using the allele frequency net database, which estimates population coverage to gauge the likelihood of an immune response in specific groups. As a result, in Russia’s Chuvash population, where the HLA-DRB*101:01 allele frequency is 0.18, an estimated 18% may respond to the vaccine. Similarly, in Spain’s Pas Valley, the HLA-DRB*115:01 frequency of 0.32 suggests 32% coverage. However, frequency data for alleles DRB3*02:02, DRB5*01:01, and DRB3*01:01 were unavailable in the database, highlighting a limitation in the current analysis (Table 4).

Table 4.

Estimated population coverage (%) based on the HLA allele frequencies for selected regions

Allele	Population	Allele frequency	Sample size	Coverage
DRB1*01:01	Russia, Chuvash	0.18	82	18
DRB1*15:01	Spain, Pas Valley	0.32	88	32

Multiepitope subunit vaccine design

The final vaccine design included 12 CTL epitopes, 6 HTL epitopes, and 6 B-cell epitopes, all of which met the criteria for binding affinity, antigenicity, nontoxicity, and the absence of allergenicity. AAY linkers were used to connect CTL epitopes, whereas GPGPG linkers joined B-cell and HTL epitopes. Griselimycin (APD ID: AP02688) was added as an adjuvant to both the N- and C-termini via EAAAK linkers to potentiate immunogenicity. The PADRE sequence (AKVAAWTLKAAAC) was also included to enhance the immune response. The final vaccine construct comprised 380 amino acids. The construct was eventually checked for satisfying all the criteria, such as, solubility, antigenicity, nontoxicity, and the absence of allergenicity. The outline of this structure is illustrated in Fig. 1.

Fig. 1.

Illustration of the multiepitope vaccine construct. The epitopes were constructed using amino acids and including adjuvants at both the N- and C-termini connected with the first CTL epitope via an EAAAK linker. The CTL epitopes are linked with AAY linkers, whereas the HTL and B-cell epitopes are linked with GPGPG linkers

Antigenicity and allergenicity of vaccine candidate

The six target proteins were evaluated via the VaxiJen v2.0 web tool, and all the proteins displayed high antigenicity, with scores ranging from 0.61–0.81.The criterion for antigenicity was set to default threshold of 0.4. The antigens IutA, FimH, ireA, IroN, FyuA and ChuA displayed antigenicity scores of 0.61, 0.67, 0.67, 0.807, 0.65 and 0.64, respectively. The vaccine construct had an antigenicity score of 0.73. Predictions from AllerTOP v.2 and AllergenFP revealed that the construct was not allergenic.

Physicochemical properties, solubility, and secondary structure of vaccine candidate

The molecular weight and isoelectric point (pI) of the vaccine construct were predicted to be 53.19 kDa and 9.27, respectively. The predicted half-life was 30 hrs in mammalian reticulocytes, more than 10 hrs in E. coli, and more than 20 hrs in yeast. The instability index (II) of 35.07 suggests that the construct is highly stable. Additionally, the aliphatic index score of 65.58 indicates thermostability, and the grand average hydropathicity (GRAVY) of − 0.337 demonstrates that the protein is hydrophilic. A solubility score of 0.44 predicts high solubility upon expression.

Overall, the PROTEUS2 protocol predicted that the vaccine construct would consist of 22% α-helices, 18% β-sheets, and 60% coil content. For solvent accessibility, 20% of the residues were predicted to be medium-exposed, 53% fully exposed, and 26% buried.

Tertiary structure and stability

The tertiary structure of the vaccine was modeled via the I-TASSER server, which generated several structures with an estimated root-mean-square deviation (RMSD) of 11.2 ± 4.6 Å. Refinement of the crude protein was performed via GalaxyRefine, resulting in five refined models. Model 3 was selected for further analysis on the basis of its quality, which included Global Distance Test - High Accuracy metric as GDT-HA (0.9437), RMSD (0.424), MolProbity (2.296), the clash value (16.6), the low rotamer value (0.5), and the Ramachandran favored value (89%) (Suppl. Figure 1 [A]). ERRAT analysis revealed that the model had an overall quality score of 84.9%, and a ProSA-web Z score of − 5.98 was obtained (Suppl. Figure 1B).

The refined structure was then evaluated via the Ramachandran plot analysis, which revealed that 70.3% of the residues were in favored regions, 24.9% were in generously allowed regions, 3.5% were in additional allowed regions, and 1.2% were in disallowed regions (Supple. Figure 1C).

The ClusPro 2.0 and LIGPROT v2.2 software enabled docking and analysis of possible interactions between the TLR4 receptor ( and the multiepitope vaccine. The docked structure was selected based on the minimum binding energy value of − 43.71 kcal/mol (Fig. 2A).

Fig. 2.

A Visualization of the interaction interface between the multiepitope vaccine construct (top molecule) and the TLR4 receptor (bottom molecule) using the Rosetta protocol, highlighted with a red surface using the ClusPro algorithm. B 3D molecular alignment of the top 10 multiepitope construct-TLR4 complexes predicted with the Rosetta protocol, with the reference structure depicted in green for comparison. C Scatter plot showing the correlation between energy, measured in Rosetta energy units (REUs), and root-mean-square deviation (RMSD), illustrating the stability and energy distribution of the complexes

The interaction interface between the multiepitope construct and the TLR4 receptor generated 19 hydrogen bonds, and 29 and 21 hydrophobic interactions for the ligand and the receptor, respectively. Stability analysis using the -minimum-energy complex structure as a reference was performed in Rosetta revealing a similarity in the top ten conformations (Fig. 2B) and a correlation between the RMSD values and the conformational energy of the complex, with an r² value of 0.71. A confidence interval (CI) multiplier of 1.0 was used to exclude noisy data as outliers instead of 3.91, which corresponds to a much broader range, where 95% of the data lies within ± 1.96 standard deviations from the mean for a more realistic representation of the data variability (Fig. 2C).

Molecular dynamics simulation

The MD simulation data, complemented by MM/GBSA binding free energy analysis, provides molecular insights into the structural stability and interaction dynamics of the TLR4-vaccine complex over a 0.4 µs period. The root mean square deviation (RMSD) demonstrated that the TLR4 component maintained moderate fluctuations (4.0 Å), while the vaccine component initially showed a significant increase in RMSD (11.0 Å), stabilizing after approximately 0.1 µs, reflecting the initial conformational adjustment of the vaccine before achieving a steady binding state with TLR4 (Fig. 3A).

Fig. 3.

RMSD (A), RMSF (B), Rg (C), and the fraction of the dominated intermolecular H-bonds calculated for the TLR4-vaccine complex during 0.4 μs MD simulation (D)

Conversely, the root mean square fluctuation (RMSF) per residue for TLR4 exhibited low fluctuations compared with higher RMSF values in specific regions of the vaccine, suggesting flexible regions within the vaccine structure. This flexibility potentially enhances adaptability and interaction with TLR4 without compromising the complex’s stability. The RMSF analysis provides valuable insight into the flexibility of specific residues within the TLR4-vaccine complex, which can be correlated with the stability and preservation of secondary structures—alpha helices, beta sheets, and coils—across multiple MD models.

Residues with low RMSF values typically correspond to more rigid, well-structured regions, such as alpha helices and beta sheets. These elements form the stable core of protein-protein interfaces, contributing significantly to the binding stability observed in the TLR4-vaccine complex. The consistent alignment across MD models, as depicted in Fig. 3B, suggests that these secondary structures are not only well-preserved but also contribute to the robust binding conformation between TLR4 and the vaccine construct.

In contrast, regions with higher RMSF values, which often correspond to loop regions or flexible coils, and protein terminal parts might be more adaptable and may allow the vaccine construct to accommodate the TLR4 binding surface without compromising the stability of the core interaction. This flexibility at certain interface points might increase binding adaptability, as these regions can adjust conformation slightly to maintain optimal contact with TLR4 while allowing the structured regions (alpha helices and beta sheets) to stabilize the interaction. The combination of stable, low-RMSF regions and adaptable high-RMSF loops likely contributes to the complex’s overall stability and potential to elicit an effective immune response.

The radius of gyration (Rg) values for both TLR4 and the vaccine remain stable, indicating that the complex maintains its compact structural integrity, which is essential for consistent binding interactions throughout the simulation (Fig. 3C). The fraction of dominant intermolecular hydrogen bonds (H-bond fraction > 0.5) provided significant contributions from bonds, such as Asp1013/Arg208, Leu983/Arg356, and Asn239/Lys1099, collectively reinforcing the stability of the TLR4-vaccine interface (Fig. 3D).

MM/GBSA binding free (ΔG_gbsa) energy analysis revealed a refined and favorable binding energy of − 41.83 kcal/mol for the complex, with strong contributions from vdW and electrostatic energy (Table 5). This energetically favorable profile is also influenced by solvation effects, where the positive polar solvation energy is partially offset by the nonpolar solvation energy. Collectively, the MD and MM/GBSA findings highlight the adaptability of the vaccine construct and its energetically favorable interaction with TLR4 fragment.

Table 5.

Different energy terms in kcal/mol for the vaccine construct predicted by the MM-GBSA method

Energy term	Mean	Std. Dev.	Std. Err. of Mean
Evdw	− 189.03	12.49	1.76
Eel	− 3539.53	129.18	10.27
Egb	3712.13	127.88	18.09
Esurf	− 25.4	1.82	0.26
ΔGgas	− 3728.56	130.72	18.49
ΔGsolv	3686.73	126.98	17.96
ΔGgbsa	− 41.83	10.4	1.47

The abbreviations E_vdw, E_el, E_gb, E_surf, ΔG_gas, ΔG_solv, and ΔG_gbsa correspond to the van der Waals energy, electrostatic energy, polar solvation energy, nonpolar solvation energy, gas phase energy, net solvation energy, and net MM/GBSA system energy, respectively

In silico cloning

The optimized DNA sequence produced by NovoPro presented favorable expression parameters, with a CAI of 0.79 and a GC content of 54.85%. The adapted codon sequence was cloned and inserted into the pET28a(+) vector via SnapGene software, with EcoRV and Eco53kI restriction sites flanking the multiepitope sequence for appropriate integration into the plasmid (Fig. 4).

Fig. 4.

In silico cloning. Multiple epitope antigens (red) were integrated into the pET28a (+) expression vector between the Eco53kI and EcoRV restriction sites

Discussion

Many microorganisms present challenges for vaccine development, especially those with complex cultivation and attenuation needs, necessitating the adoption of new technological approaches (Poland et al. 2009). In recent years, multiomics and immunoinformatics-guided vaccine development strategies have taken precedence over traditional methods (Hasan et al. 2019). Multiepitope peptide vaccines targeting bacterial virulence proteins, such as those from UPEC, offer promising options for effective vaccine design.

The use of immunoinformatics has facilitated the creation of potent multiepitope vaccines against several pathogens, such as Acinetobacter baumannii (Dey et al. 2023), Streptococcus pneumoniae (Shafaghi et al. 2023), Staphylococcus aureus (ul Qamar et al. 2021), monkey pox virus (Tan et al. 2023), P. falciparum (Maharaj et al. 2021) and SARS-CoV-2 (Umitaibatin et al. 2023; Sarvmeili et al. 2024).

These vaccines are engineered to elicit strong B- and T-cell responses, with high population coverage, antigenicity, and safety profiles. However, experimental validation is necessary to verify the safety and effectiveness of these computationally designed vaccines before they are tested in clinical trials (Shawan et al. 2023).

Vaccine development for E. coli pathogens presents unique challenges because of the genetic plasticity and adaptability of this bacterium. In addition, the lack of universally applicable testing methods for disease incidence in remote regions and the presence of multiple bacterial pathotypes complicate vaccine trials (Fleckenstein 2021). Despite these challenges, the multiepitope vaccine proposed in this study, which targets conserved virulence proteins, has the potential to be effective against various E. coli pathotypes.

Here we have proposed the construction of a novel chimeric vaccine candidate, considering a virulence protein source that exhibits a high level of conservation in most common E. coli pathogens. Consequently, the envisioned vaccine could exhibit efficacy against various E. coli pathogens. Moreover, the proteins used for the vaccine design are located in the outer membrane, playing a critical role in the survival of bacteria. With a notable antigenicity score, they demonstrated low similarity to human proteins, thus guaranteeing their suitability for use in clinical trials.

The virulence proteins offered a large array of epitopes (255), which were predicted on the basis of a set of HLAs presenting wide population coverage (> 97% for MHC class I and > 99% for MHC class II). Our selection criteria allowed us to reach a final set of epitopes, CTLs, and HTLs, which have high affinity scores for MHC molecules, and being recognized by B cells. This feature is indispensable to vaccinology because of its relevance in neutralizing antibodies against the pathogen and prolonged antibody production by T cells thus facilitating the activation of the CTL response (Chaplin 2010; Naz et al. 2020; Ullah et al. 2020).

The vaccine selection criteria focused on epitopes with high antigenicity and population coverage, indicating the potential for broad immune responses. The adjuvant is employed substantially for the construction of multiepitope vaccine (Guo et al. 2014; Ojha et al. 2020) and is linked to triggering the host’s immune response in experiments conducted in vivo (Li et al. 2016; Hou et al. 2014; Kalenik et al. 2018). The use of linkers, such as EAAAK, AAY, and GPGPG (Solanki et al. 2019; Shey et al. 2019; Athanasiou et al. 2017; Nezafat et al. 2014; Safavi et al. 2019), enhanced antigen processing and presentation (Athanasiou et al. 2017). The inclusion of the PADRE sequence minimized HLA polymorphism issues, improving vaccine efficacy across different populations (Ghaffari-Nazari et al. 2015).

Furthermore, the selected epitopes were all predicted as possible antigens, illuminating the immunogenic potential of these sequences. The multi-epitope vaccine exhibited multiple intermolecular connections with human TLR4, including hydrogen bonds and hydrophobic interactions through protein-protein docking. The physicochemical properties of the multiepitope candidate strongly indicate the feasibility of heterologous expression and antigen purification. The projected stability (instability index of 35.07 and aliphatic index of 65.58) and half-life in E. coli (10 hrs ) suggest the use of heterologous expression of this antigenic vaccine in this host. This aliphatic index reflects the relative volume occupied by aliphatic amino acids in the side chains (Sarkar et al. 2020; Kar et al. 2020). The validation of the 3D structures of the constructed vaccines via ERRAT and Ramachandran plot analysis generated a good quality factor and a high percentage of favored regions. Thus optimization of the vaccine candidate codon was executed according to E. coli strain K12 and in silico cloning into a common expression vector, pET28a (+), was performed. In silico analyses, including molecular docking and dynamics simulations, revealed the strong binding affinity of the vaccine for TLR4 and its potential to activate immune responses against UPEC. The physicochemical properties of the vaccine further suggest its suitability for heterologous expression and antigen purification.

Conclusion

This study presents an optimized immunoinformatic approach for designing a multiepitope peptide vaccine targeting key virulence factors of UPEC, a significant causative agent of UTI. By leveraging advanced computational tools, we identified CTLs, HTLs, and B-cell epitopes with high antigenicity, and the absence of allergenicity, and toxicity. The final vaccine construct demonstrated strong binding affinity and stable interactions with TLR4, suggesting its potential to stimulate both cellular and humoral immune responses. MD simulations and protein-protein affinity calculations analyzed the interaction profile of the vaccine-TLR4 complex, highlighting its promise for effective immune activation. The predicted physicochemical properties of the vaccine candidate further support its feasibility for expression in E. coli, making it suitable for laboratory production and downstream testing.

While these in silico findings are encouraging, experimental validation through in vitro and in vivo studies is essential to confirm the vaccine’s efficacy, immunogenicity, and safety. Such investigations will help bridge the gap between computational predictions and clinical applications. Ultimately, this multiepitope vaccine candidate holds promise as a preventive strategy against recurrent UTIs caused by UPEC, offering a potential solution to combat increasing antibiotic resistance in UTI pathogens.

Author contributions

Hina Khalid and Sergey Shityakov conceived the study. Hina Khalid performed the immunoinformatics analysis, including epitope mapping, antigenicity assessments, and population coverage analysis. Sergey Shityakov performed the molecular dynamics simulations, molecular docking, and the structural validation. Hina Khalid wrote the initial manuscript draft. Sergey Shityakov prepared Figs. 1, 2 and 3 and contributed to the editing and review of the manuscript. Both authors reviewed and approved the final version of the manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.