Progress toward a vaccine for extraintestinal pathogenic E. coli (ExPEC) II: efficacy of a toxin-autotransporter dual antigen approach

ABSTRACT

Extraintestinal pathogenic Escherichia coli (ExPEC) is a leading cause of worldwide morbidity and mortality, the top cause of antimicrobial-resistant (AMR) infections, and the most frequent cause of life-threatening sepsis and urinary tract infections (UTI) in adults. The development of an effective and universal vaccine is complicated by this pathogen’s pan-genome, its ability to mix and match virulence factors and AMR genes via horizontal gene transfer, an inability to decipher commensal from pathogens, and its intimate association and co-evolution with mammals. Using a pan virulome analysis of >20,000 sequenced E. coli strains, we identified the secreted cytolysin α-hemolysin (HlyA) as a high priority target for vaccine exploration studies. We demonstrate that a catalytically inactive pure form of HlyA, expressed in an autologous host using its own secretion system, is highly immunogenic in a murine host, protects against several forms of ExPEC infection (including lethal bacteremia), and significantly lowers bacterial burdens in multiple organ systems. Interestingly, the combination of a previously reported autotransporter (SinH) with HlyA was notably effective, inducing near complete protection against lethal challenge, including commonly used infection strains ST73 (CFT073) and ST95 (UTI89), as well as a mixture of 10 of the most highly virulent sequence types and strains from our clinical collection. Both HlyA and HlyA-SinH combinations also afforded some protection against UTI89 colonization in a murine UTI model. These findings suggest recombinant, inactive hemolysin and/or its combination with SinH warrant investigation in the development of an E. coli vaccine against invasive disease.

INTRODUCTION

Extraintestinal pathogenic Escherichia coli (ExPEC) represents the most prevalent Gram-negative bacterial pathogen and is a primary contributor to mortality due to antimicrobial resistance (AMR) globally (both deaths attributable to and associated with AMR) (1, 2). ExPEC comprises the pathotypes of uropathogenic E. coli (UPEC), neonatal meningitis E. coli, and septicemia-associated isolates (3). ExPEC is the primary cause of bacteremia and urinary tract infections (UTIs) and a frequent cause of neonatal meningitis (4, 5). In the United States, over 970,000 sepsis cases are admitted annually, with an 8.7% yearly increase in incidence among hospitalized patients, accounting for over 50% of hospital deaths (6, 7). Based on the Centers for Disease Control and Prevention (CDC) multiple cause-of-death data (1999–2014), 6% of all deaths involved sepsis, 22% of these cases listing sepsis as the underlying cause (8). Moreover, in 2017, approximately 48.9 million new cases of sepsis were recorded globally, with 11 million sepsis-related deaths reported, accounting for 19.7% of all worldwide deaths (9). In addition, sepsis management remains a major challenge for healthcare systems worldwide, resulting in a disproportionately high burden in terms of cost and hospital resource utilization. In the United States, sepsis management costs surpass those for any other disease, exceeding $24 billion in 2013, representing 13% of total hospital expenses and growing at three times the rate of other admissions (10). E. coli has emerged as the predominant causative agent of bloodstream infections (BSIs) in both community and hospital settings over the past decade, accounting for 27.1% of all bacteremia cases. Moreover, the incidence rate of E. coli bacteremia is estimated at 48 per 100,000 person-years, exhibiting a notable increase with advancing age (11).

Next to their virulence, the second concerning feature of ExPEC is they are a leading cause of AMR, which frequently results in treatment failure, increased hospitalization rates, and exacerbated morbidity and mortality. A recent review assessing the global burden of bacterial AMR across 204 countries and territories identified antibiotic-resistant pathogenic E. coli as a primary cause of mortality associated with drug resistance, accounting for approximately 200,000 deaths due to antimicrobial-resistant E. coli and around 800,000 deaths linked to AMR E. coli in 2019 (2). The CDC reports that over two million people in the United States contract antibiotic-resistant diseases annually, with AMR contributing an additional $20 billion to direct healthcare costs and roughly $35 billion in lost productivity each year (12). Furthermore, other studies found that antimicrobial-resistant ExPEC infections impair the capacity of the immune system to clear infections, including complications in patients following prostate biopsy (13), solid organ transplant (14, 15), or undergoing chemotherapy, dialysis, surgery, and joint replacement (12). The recent emergence of a sequence type (ST) 131 (ST131), which combines both pan resistance with high levels of virulence, has become globally disseminated (16). Furthermore, ST131 E. coli isolates maintain a balance between colonization, virulence, and antibiotic resistance without incurring a fitness cost, attributed to their distinct virulence profiles and expanded number of virulence genes compared to non-ST131 isolates (17, 18). In the United States, ST131 significantly contributes to the resistance of clinical E. coli isolates, accounting for approximately 70% of fluoroquinolone-resistant strains and over 50% of MDR isolates (19–21). Other sequence types, such as ST95 and 73, also remain prevalent and well-recognized as highly pathogenic ExPEC strains among clinical isolates in patients with UTIs and BSIs (22, 23).

A vaccine against ExPEC is expected to reduce morbidity and mortality and help mitigate the AMR crisis. Over the past few decades, numerous groups have pursued protective immunity against ExPEC through various strategies, including the use of inactivated bacteria (24–28), bacterial lysates (29, 30), or O-specific polysaccharide (O-antigen) conjugate vaccines (31–33). Also, flagellin protein-based vaccines such as FliC (34–36) or fimbrial-based vaccines such as FimH (from type 1 fimbriae) (37) and other fimbrial or non-fimbrial-based vaccines, such as adhesin FdeC (38), PapG fimbrial adhesin (39), and Dr fimbriae (40) have also been studied. Furthermore, proteins involved in iron acquisition (IroN, lutA, IreA, FyuA, and siderophores) (41–49) and toxin-based products such as insoluble α-hemolysin or CNF1 (cytotoxic necrotizing factor 1) (50–52) have also been considered. Moreover, in a series of clinical trials, researchers investigated the efficacy and safety of SolcoUrovac (a whole-cell vaccine comprising heat-killed bacteria from 10 uropathogenic strains) or Uro-Vaxom (OM-89, a bacterial extract prepared from 18 uropathogenic E. coli strains) in vaginal mucosal immunization against recurrent UTIs and demonstrated the beneficial effects in women (53–57). Another clinical trial example is ExPEC4V, a tetravalent O-polysaccharide conjugate vaccine, has elicited functional antibody responses and is currently in the phase 2 randomized controlled trial (58–63). Additionally, the ExPEC10V bioconjugate vaccine also demonstrated both tolerability and immunogenicity in elderly adults (64). However, the high heterogeneity of O-specific polysaccharides may limit the development of a polysaccharide vaccine capable of preventing all ExPEC infections (59). At the moment, no ExPEC vaccine has been approved by the U.S. Food and Drug Administration.

The α-hemolysin (HlyA) is a critical and commonly detected secreted cytotoxic virulence factor in ExPEC, associated with upper UTIs such as cystitis or pyelonephritis (65). HlyA, a pore-forming toxin belonging to the RTX toxin family (repeats in toxin), exhibits cytotoxic activity against various species and cell types, potentially causing severe tissue damage (66). Additionally, HlyA can lyse erythrocytes and damage effector immune cells at high concentrations (67, 68), promote bladder epithelial cell exfoliation, and induce apoptosis in target host cells at low concentrations (69). Clinically, HlyA is linked to severe UTIs that can lead to renal complications and permanent renal scarring (70, 71) and may also cause endothelial damage and renal vasoconstriction (72). The hlyA gene exhibits high prevalence in clinical E. coli isolates from patients with bloodstream infections (73–75), UTIs (76–78), and pregnant women, reaching up to 61.5% in E. coli strains causing cystitis and 78.6% in strains causing pyelonephritis (79). Furthermore, an analysis of the hlyA gene distribution among major ExPEC clones revealed its prevalence was significantly higher in ST73 (64.6%) compared to ST131 (14.8%) and ST95 (13.5%) and was more frequently found in strains from phylogroup clades B and C (80, 81).

Considering the significance of HlyA in ExPEC pathogenesis and the high prevalence of hlyA sequence in clinical E. coli isolates, we sought to investigate the potential of pro-HlyA, the inactive precursor form of HlyA (82), as a vaccine candidate against ExPEC infections, with and without a previously reported autotransporter antigen showed to be highly protective (83). Here, we present data suggesting that both pro-HlyA and a pro-HlyA/SinH combinatorial mixture generates robust protection against various virulent and concerning sequence types of ExPEC strains in multiple murine infection models.

MATERIALS AND METHODS

Bacterial strains and culture conditions

The E. coli strains utilized in this study were obtained from a single colony grown on the Lysogeny Broth (LB) plate [10 g/L tryptone, 0.5 g/L sodium chloride (NaCl), and 5 g/L yeast extract]. Bacterial cultures were incubated at 37°C after resuscitation from a frozen stock. The ExPEC ST131 strains used in the study, JJ1886, JJ1901, JJ2050, JJ2528, and JJ2547, were kindly provided by James R. Johnson (84). UPEC strains UTI89 (O18:K1:H7, ST95) (85) and CFT073 (O6:K2:H1; ST73) (86) were kindly provided by Kathryn Patras. E. coli strains W0008 (ST127-like), W0044 (ST405-like), W0128 (ST648-like) were isolated from the blood or feces of hospitalized patients with bacteremia. The number of colony-forming units (CFU) administered was determined by correlating the optical density (OD) at 600 nm to the number of colonies observed after plating.

The hlyA sequence distribution and HlyA alignment

Complete E. coli genomes were obtained from NCBI’s RefSeq database (87) and sorted into phylogroups as previously reported (81). The sorted genomes were then categorized by sequence types using the MLST software (https://github.com/tseemann/mlst) which uses the PubMLST databases (https://pubmlst.org/) (88). The categorized genome database was then used as a custom BLAST (version 2.8.1) (89–91) database to search for hits to the hlyA, hlyB, hlyC, and hlyD genes from the hly operon of E. coli UTI89 genome (accession: CP000243.1). To design Fig. 1A, the underlying phylogenetic tree diagram was created using the autoMLST software (92) in concatenated alignment mode with 1,000 UltraFast Bootstrap replicates using representative genomes from the 8 E. coli phylogroups. The representative phylogenetic diagram was then overlaid with pie charts created in GraphPad Prism using BLAST hit results. Finally, the figures were combined using Biorender. All software used default settings unless otherwise specified. Open reading frames that overlapped with BLAST hits (described above) were extracted and translated using the bacterial translation code (translation table 11) in Geneious 2023.1.1. Translated HlyA sequences were then aligned using Geneious Alignment Software with free end gaps and otherwise default settings after truncated ORFs were removed. The resulting alignment was sorted by a phylogenetic tree annotation that was created using RXaML (version 8.1.1) with the GAMMA BLOSUM62 protein model and 100 Bootstrap replicates from the Rapid Bootstrap algorithm to create a consensus tree. Vibrio parahaemolyticus hemolysin A (accession: WP_041955411.1) was used as an outgroup.

Fig 1.

The ExPEC-associated genetical level of Hemolysin A, HlyA structure prediction, and the purification of pro-HlyA. An analysis of a database of 1,348 complete E. coli genomes that have been phylogenetically categorized shows that the hlyA sequence is predominantly found in ExPEC-associated sequence types of the B2 phylogroup. (A) Phylogenetic representation of hlyA sequence distribution. BLAST was used to compare the hlyA nucleotide sequence to a database of complete E. coli genomes from NCBI’s Genbank that had been sorted into phylogroups using a previously described in-house method and into sequence types using MLST software (https://github.com/tseemann/mlst). Pie charts were made using GraphPad Prism, and the final figure was created using Biorender. (B) Amino acid alignment of HlyA. Tickmarks represent disagreements with the majority consensus at that residue and are colored using the Rasmol coloring scheme. Open reading frames overlapping with BLAST hits for hlyA were translated, aligned using Geneious Alignment (Geneious 2023.1.1) and then sorted using a tree annotation. The phylogenetic tree annotation was created using RXaML (version 8.1.1) and represents a consensus tree from 100 Bootstrap replicates with Vibrio parahaemolyticus hemolysin A (accession: WP_041955411.1) used as an outgroup. (C) Overall predicted structure of HlyA. AlphaFold2 generated predicted structure shows three domains for HlyA with one linker: N-terminal helical domain (residues 1–279, red), three helix bundle (residues 321–437, blue), beta-helix C-terminal domain (residues 438–1023, green), and linker between N-terminal domain and helix-bundles (residues 280–320, gray). (D) Predicted structure of C-terminal beta helix domain. A large beta helix dominates the overall structure of this domain. Both N- and C-ends of the domain contain two beta strands and two alpha helices. (E) Plasmid pSU-hlyA (encoding the hlyA sequence) and plasmid pK184-hlyBD (encoding hlyB and hlyD sequence) were co-transformed into E. coli BL21 (DE3) cells. Bacterial cultures expressing the recombinant pro-HlyA antigen secreted the protein into the supernatant, which was subsequently harvested, filtered, and concentrated. The purified antigen was analyzed by SDS-PAGE and stained with Coomassie blue stain and silver stain. Predicted size of pro-HlyA, 110 kDa. The SDS-PAGE result was annotated using BioRender. (F) The coverage rate of pro-HlyA was determined by per-band sequencing through mass spectrometry.

Prediction of protein structure for HlyA with AlphaFold2

The nucleotide sequence of HlyA was used to recreate the translated amino acid sequence using ExPASy. All six possible reading frames (three forward, three backward) were generated, and the frame that matched the sequence for complete HlyA was used as the amino acid sequence for structure prediction. ColabFold’s AlphaFold2-Advanced Google Notebook (Google, Mountain View, CA) was used to generate predictions from amino acid sequences (93, 94). For multiple sequence alignment (MSA) necessary to build the consensus model for the structure of HlyA, we used MMseq2 (95, 96). Five prediction runs were performed, with each run using a randomly chosen initiation point for the start of prediction runs. These models were ranked using the following two metrics: (i) pLDDT (predicted LDDT-Cα) with its ability to quantify the confidence of model per residue calculated by utilizing distances between Cα atoms in multiple reference models and (ii) AlphaFold-generated PAE (Predicted Aligned Error) for every residue, a numerical value of expected position error per residue (93, 97). The model with the highest average pLDDT and lowest PAE was chosen as the best-predicted structure of HlyA, and AMBER Force Field was applied to relax the structure (98). The predicted structure was compared against the list of previously solved structures of RTX toxins deposited on PDB aligning spatial coordinates of models by domains (99, 100). Furthermore, Foldseek search was used to search for similar solved and AlphaFold-predicted structures on multiple databases (101, 102). Additionally, these structures were aligned with the predicted structure by UCSF ChimeraX’s alignment feature using the Needleman-Wunsch algorithm with BLOSUM-62 similarity matrix (103). ChimeraX was used for analyzing the structural features of the predicted model, determining local physical properties within domains, and visualizing the model.

Plasmid construction

The plasmid for the candidate vaccine antigen SinH-3 was constructed using a previously described method (83). The plasmids pSU-hlyA (encoded the candidate vaccine antigen pro-HlyA, Uniprot entry: P08715) and pK184-hlyBD (encoded the necessary transport complex components HlyB and HlyD) were kindly provided by Lutz Schmitt (82). The mRNA plasmid was constructed by cloning the SinH-3 gene from ExPEC sequence type 131 (ST131) strain JJ1887 genomic DNA (SinH-Ig-like domains-3, encoding the C-terminal passenger Ig-like domain-3 fragment of sinH, amino acid residues 602–724) and the pro-HlyA was cloned from E. coli (Uniprot entry: P08715). Both sequences (hlyA and sinH-3) were submitted to Creative Biolabs for constructing a sinH-3: hlyA mRNA construct with IL-2 signal peptide and P2A self-cleaved sequence, called pIVTScrip-mRNA-IL2-sig_hlyA-P2A-IL2-sig_sinH (hereinafter named Dual-Hit mRNA construct). The Dual-Hit mRNA constructs were further enclosed in Cationic Lipid Nanoparticle (Cationic Lipid Nanoparticle (SM-102/DSPC/Cholesterol/DMG-PEG = 50: 10: 38.5: 1.5) and stored in Tris-based buffer at –80°C.

Vaccine antigens preparation

The recombinant SinH-3 protein was expressed as fusions with glutathione-S-transferase (GST) using E. coli BL21(DE3) and purified as previously described (83). To purify the recombinant protein pro-HlyA, the plasmid pSU-hlyA, containing the C-terminal secretion signal of HlyA (Uniprot entry: P08715), and pK184-hlyBD, encoding HlyB and HlyD essential for the transport complex, were co-transformed into E. coli BL21(DE3) cells (82). A single pSU-hlyA and pK184-hlyBD co-transformed E. coli BL21(DE3) colony was used to inoculate a 300 mL flask containing 150 mL of Lysogeny broth (LB) medium and cultured overnight. The overnight culture was then used to inoculate a 2-L flask containing 800 mL of LB medium, which was grown at 37°C until it reached optical density at 600 nm (OD₆₀₀) of 0.4–0.6. Gene expression was induced with 1 mM Isopropyl β-d-1-thiogalactopyranoside (IPTG) (Sigma-Aldrich, St. Louis, MO), and the culture was incubated overnight at 37°C and 150 rpm. Following induction, the supernatant containing secreted pro-HlyA protein was collected by centrifugation [Thermo Fisher Scientific, Sorvall RC 6+, SLA-3000 (Rotor), 10,000 × g for 30 min at 4°C] and filtered through the 0.22-µm Vacuum Driven Sterile Filters (Sigma-Aldrich, St. Louis, MO). The filtered supernatant was subsequently concentrated to 1 mL using Amicon Ultra-15 Centrifugal Filter Units (Millipore Sigma, Burlington, MA) with a 100 kDa molecular-weight cut-off. All antigens were analyzed by SDS-PAGE via Coomassie Brilliant Blue staining and Silver Stain (Pierce Silver Stain Kit, Thermo Fisher Scientific, Waltham, MA), and the expression of both purified proteins was confirmed by mass spectrometry as previously described (83). Both recombinant proteins were stored at −20°C until further use and handled at 4°C. The control group consisted of culturing a single untransformed E. coli BL21(DE3) colony only (hereinafter referred to as control supernatant). The resulting supernatant was collected and concentrated using the exact same procedure as that used for the purification of pro-HlyA above, and the supernatant control was also analyzed by SDS-PAGE via Coomassie Brilliant Blue staining and Silver Stain.

Experimental animals

Experimental animals used in this study were 6- to 8-week-old BALB/cJ mice obtained from Jackson Laboratories (Bar Harbor, ME). They were provided with sterile food and water ad libitum and housed in filtered cages with 3–4 mice per cage. All experimental procedures performed on mice were approved in accordance with relevant guidelines and regulations from “The Guide and Care and Use of Laboratory Animals” (National Institute of Health) and were approved by Baylor College of Medicine’s Institutional Animal Care and Use Committee under protocol number AN-5177.

Vaccination

For experiments involving protein-subunit vaccines, purified proteins were mixed with alum adjuvant (G-Bioscience, St. Louis, MO) in a 2:1 ratio of antigen to adjuvant, following the manufacturer’s guidelines. Female BALB/cJ mice (6- to 8-week-old) were given three subcutaneous injections (S.C) of either 50 µg of pro-HlyA or a mixture of SinH-3 and pro-HlyA (50 µg each) on days 0, 14, and 28. Control groups were either given vaccinations supernatant control samples [comprising the same volume of control supernatant (30 µL), described above, mixed with alum adjuvant (30 µL) (Fig. 2B, C and 3B, C) or left unvaccinated (Fig. 2E, F, G, 3E, F, G, 4B, C, D, 5B, C and 6B through D]. In experiments involving mRNA vaccines, 6-week-old male BALB/cJ mice were given three intramuscular injections (I.M) of either 2 µg Dual-Hit mRNA construct (low-dose group, 40 µL to one hind leg muscle), 5 µg Dual-Hit mRNA construct (high-dose group, 40 µL to one hind leg muscle) (104), or 50 µL of Tris-based buffer (control group).

Fig 2.

Evaluating the protective efficacy of pro-HlyA against UTI89 infections in the murine model of bacteremia and mortality. (A) The scheme of the murine bacteremia model using UTI89. Female BALB/cJ mice were subcutaneously immunized with either pro-HlyA (N = 16) or control supernatant (N = 16), followed by an intraperitoneal (I.P.) injection of 1 × 10⁸ CFU of UTI89. After 16 h of infection, organs (kidney, spleen, liver) were harvested, homogenized, and plated to determine bacterial loads (CFU/mL). (B) Scatter plot with bar representing total UTI89 bacterial dissemination combining counts from all organs; (C) or the organ-specific UTI89 bacterial dissemination in each organ type post-necropsy. (D) The scheme of the murine mortality model using UTI89. Female BALB/cJ mice were subcutaneously immunized with pro-HlyA (N = 12) or left unvaccinated (N = 8), followed by an intraperitoneal (I.P.) injection of 5 × 10⁷ CFU of UTI89. Mice were monitored twice daily for 10 days. The moribund or deceased mice were euthanized and necropsied to determine bacterial levels in organs (CFU/mL). Four moribund vaccinated mice were euthanized at 3 d.p.i., forming the 3 d.p.i. group (N = 4). All remaining survivors were euthanized at 10 d.p.i., forming the 10 d.p.i. group (N = 8). (E) The survival rate of pro-HlyA immunized mice after UTI89 infection was assessed using the Gehan-Breslow-Wilcoxon comparison. (F) Scatter plot with bar representing the total UTI89 bacterial dissemination combining counts from all organs at 3 d.p.i. and 10 d.p.i.; (G) or the organ-specific UTI89 bacterial dissemination in each organ type (combining counts from 3 d.p.i. and 10 d.p.i.) post-necropsy. Schemes were created in BioRender. Scatter plots with bars and Kaplan Meier survival curves were exported from Graphpad Prism and annotated using BioRender.

Fig 3.

Evaluating the protective efficacy of Dual-Hit against UTI89 infections in the murine model of bacteremia and mortality. (A) The scheme of the murine bacteremia model using UTI89. Female BALB/cJ mice were subcutaneously immunized with either Dual-Hit (N = 15) or control supernatant (N = 16), followed by an intraperitoneal (I.P.) injection of 1 × 10⁸ CFU of UTI89. After 16 h post-infection, organs (kidney, spleen, liver) were harvested, homogenized, and plated to determine bacterial loads (CFU/mL). (B) Scatter plot with bar representing total UTI89 bacterial dissemination combining counts from all organs; (C) or the organ-specific UTI89 bacterial dissemination in each organ type post-necropsy. (D) The scheme of the murine mortality model using UTI89. Female BALB/cJ mice were subcutaneously immunized with Dual-Hit (N = 18) or left unvaccinated (N = 7), followed by an intraperitoneal (I.P.) injection of 5 × 10⁷ CFU of UTI89. Mice were monitored twice daily for 10 days. The moribund or deceased mice were euthanized and necropsied to determine bacterial levels in organs (CFU/mL). Six moribund vaccinated mice were euthanized at 3 d.p.i., forming the 3 d.p.i. group (N = 6). All remaining survivors were euthanized at 10 d.p.i., forming the 10 d.p.i. group (N = 12). (E) The survival rate of Dual-Hit immunized mice after UTI89 infection was assessed using the Gehan-Breslow-Wilcoxon comparison. (F) Scatter plot with bar representing the total UTI89 bacterial dissemination combining counts from all organs at 3 d.p.i. and 10 d.p.i.; (G) or the organ-specific UTI89 bacterial dissemination in each organ type (combining counts from 3 d.p.i. and 10 d.p.i.) post-necropsy. Schemes were created in BioRender. Scatter plots with bars and Kaplan Meier survival curves were exported from Graphpad Prism and annotated using BioRender.

Fig 4.

Assessing the protective efficacy of pro-HlyA and Dual-Hit against CFT073 infections in the murine model of mortality. (A) The scheme of the murine mortality model using CFT073. Female BALB/cJ mice were subcutaneously immunized with pro-HlyA (N = 12), Dual-Hit (N = 12), or left unvaccinated (N = 12), followed by an intraperitoneal (I.P.) injection of 1 × 10⁸ CFU of CFT073. Mice were monitored twice daily for 10 days. The moribund or deceased mice were euthanized and necropsied to determine bacterial levels in organs (kidney, spleen, liver, CFU/mL). Vaccinated mice that died of infection within 2 d.p.i. formed the 2 d.p.i. group for each vaccinated cohort (pro-HlyA, N = 9; Dual-Hit, N = 7). All remaining survivors were euthanized at 10 d.p.i., forming the 10 d.p.i. group for each vaccinated cohort (pro-HlyA, N = 3; Dual-Hit, N = 5). (B) Survival rates of pro-HlyA or Dual-Hit immunized mice following CFT073 infection were analyzed using the Gehan-Breslow-Wilcoxon comparison. (C) Scatter plot with bar representing the total CFT073 bacterial dissemination combining counts from all organs at 2 d.p.i. and 10 d.p.i.; (D) or the organ-specific CFT073 bacterial dissemination in each organ type (combining counts from 2 d.p.i. and 10 d.p.i.) post-necropsy. Schemes were created in BioRender. Scatter plots with bars and Kaplan Meier survival curves were exported from Graphpad Prism and annotated using BioRender.

Fig 5.

Evaluating the protective efficacy of pro-HlyA and Dual-Hit against UTI89 or CFT073 in the murine model of UTI. (A) The scheme of the murine UTI model using UTI89 or CFT073. Female BALB/cJ mice were subcutaneously immunized with pro-HlyA (N = 12/N = 12), Dual-Hit (N = 12/N = 12), or remained unvaccinated (N = 8/N = 12), followed by a transurethral injection of 1 × 10⁸ CFU of UTI89 or CFT073. After 72 h post-infection, bladders were harvested, homogenized, and plated to determine bacterial loads (CFU/mL). (B) Scatter plot with bar representing bladder UTI89 bacterial dissemination; (C) or scatter plot with bar representing bladder CFT073 bacterial dissemination.

Fig 6.

Evaluating the protective efficacy of Dual-Hit against a mixture of 10 typical ExPEC strains infection in the murine model of mortality. (A) The scheme of the murine mortality model using a mixture of ten ExPEC strains. Female BALB/cJ mice were subcutaneously immunized with Dual-Hit (N = 22), or left unvaccinated (N = 24), followed by an intraperitoneal (I.P.) injection of 1 × 10⁸ CFU (in total) of a mixture of 10 typical ExPEC strains. Mice were monitored twice daily for 10 days. The moribund or deceased mice were euthanized and necropsied to determine bacterial levels in organs (kidney, spleen, liver, CFU/mL). Two vaccinated mice that died of infection within 1 d.p.i. (N = 2) and four moribund vaccinated mice were euthanized at 3 d.p.i. (N = 4) formed the 3 d.p.i. group. All remaining survivors were euthanized at 10 d.p.i., forming the 10 d.p.i. group (N = 16). (B) Survival rates of Dual-Hit immunized mice following a mixture of 10 typical ExPEC strains infection were analyzed using the Gehan-Breslow-Wilcoxon comparison. (C) Scatter plot with bar representing the total bacterial dissemination of mixture of 10 typical ExPEC strains combining counts from all organs at 3 d.p.i. and 10 d.p.i.; (D) or the organ-specific bacterial dissemination of mixture of 10 typical ExPEC strains in each organ type (combining counts from 3 d.p.i. and 10 d.p.i.) post-necropsy. Schemes were created in BioRender. Scatter plots with bars and Kaplan Meier survival curves were exported from Graphpad Prism and annotated using BioRender.

Murine model of bacteremia

E. coli strains UTI89 were cultured under specified conditions 1 day prior to injection. On the day of injection (day 42), the strains were subcultured in LB broth at a ratio of 1:100 to an OD₆₀₀ of approximately 0.6 (Log phase, ~1 × 10⁸ CFU/mL), harvested by centrifugation (3,500 × g for 20 min at 4°C, Centrifuge 5702 R, Eppendorf North America, Framingham, MA), and suspended in an equivalent amount of 1 × PBS. Mice were intraperitoneally injected with 50 µL of the E. coli strain suspension (1 × 10⁸ CFU) on day 42, and the inoculum was quantified by plating dilutions onto LB agar. After 16 h, mice were euthanized and necropsied to collect their kidney, spleen, and liver. The organs were homogenized in 1 mL 1 × PBS using a BeadBlaster Refrigerated Homogenizer (Benchmark Scientific Inc, Sayreville, NJ, USA), and the organ homogenates were plated on LB agar plates and incubated at 37°C to determine the number of bacteria or CFU per milliliter (mL).

Murine model of mortality

On the day before injection, E. coli strains UTI89 and CFT073 were grown under the specified conditions. On the day of injection (day 42), mice were intraperitoneally injected with 50 µL of either UTI89 (5 × 10⁷ CFU) or CFT073 (1 × 10⁸ CFU) suspension. Mice were closely monitored twice daily for 10 days to observe morbidity and mortality. Survival data were collected over time, and moribund or dead animals were euthanized and necropsied to determine bacterial levels in their kidney, spleen, and liver. The organs were homogenized and plated as described above. Moribundity was determined based on multiple observable features and the scoring system listed in the animal protocol (AN-5177), including rough coat, hunched posture, lethargy, and hyperpnea. The measurement of health scores and the scoring system were described in our previous paper (105).

Murine model of urinary tract infection

UPEC strains UTI89 and CFT073 were grown and prepared as previously described. On day 42, mice were transurethrally inoculated with 50 µL of a UPEC strain suspension (1 × 10⁸ CFU) as described (106). The inoculum was quantified by plating dilutions onto LB agar. After 72 h, mice were euthanized and necropsied to collect bladders. The bladders were homogenized in 500 µL 1 × PBS and plated as described above.

Murine model of mortality from mixture of 10 strains

Ten E. coli strains, representing typical sequence types (STs) of ExPEC, were grown and prepared as previously described. On day 42, mice were intraperitoneally injected with 50 µL of a mixture of 10 ExPEC strains (equally mixed, a total of 1 × 10⁸ CFU). The inoculum was quantified by plating dilutions on LB agar. Mice were monitored twice daily for 10 days to observe their survival. Survival data were collected over time, and moribund or dead mice were euthanized and necropsied to determine bacterial levels in their kidney, spleen, and liver. The organs were homogenized in 1 mL 1 × PBS and plated as described above. Moribundity was determined by observing multiple features, including rough coat, hunched posture, lethargy, and hyperpnea.

Murine model of gastrointestinal tract colonization

Female BALB/cJ mice were subjected to a vaccination protocol, receiving three subcutaneous injections of either pro-HlyA or Dual-Hit antigens on days 0, 14, and 28, and the control group mice were administered an equal number of control supernatant injections as described previously. Prior to bacterial challenge, all mice underwent a 12 h fasting period for them to enhance infection efficacy. On day 42, mice were orally challenged with transgenic kanamycin-resistant ExPEC strains, either UTI89 or CFT073, at a dose of 5 × 10⁸ CFU/mouse via gavage. Fecal sample collection individually commenced 24 h after infection and was repeated daily for the duration of 5 days. Fecal pellets, obtained under sterile conditions, were homogenized in 1 mL 1 × PBS. The homogenates were then serially diluted, cultured on LB Agar plate with 50 µg/mL kanamycin, incubated at 37°C overnight, and determined the number of bacteria by CFU/mL.

Statistical analyses

Graphs and statistical analyses were conducted using GraphPad Prism (GraphPad Software, San Diego, CA). Significance was determined using the Mann-Whitney U test (two groups) or Kruskal-Wallis analysis of variance (ANOVA) with Dunn’s multiple comparisons correction (more than two groups). Survival curves were compared using the Genhan-Breslow-Wilcoxon curve comparison. Statistical significance was determined if the calculated P-values were less than 0.05. The lines of all the bar graphs were at the median with 95% confidence intervals (CI). Log ⁽²⁾ = 100 CFU/mL is the lower limit of detection. If no bacterial colony is detected on the plate, the CFU count will be calibrated from 0 to 1, thereby indicating the value in logarithmic form in the figures (Log⁽¹⁾ = 0). The statistical significance is represented as one star (*) for P <  0.05, two stars (**) for P <  0.01, three stars (***) for P <  0.001, and four stars (****) for P <  0.0001. The box-and-whisker plots and Kaplan Meier survival curves were generated using GraphPad Prism and annotated with BioRender.

RESULTS

Comparative genomics of hemolysin in ExPEC

Our vaccine search efforts center around a strategy to use comparative pathogenomics combined with functional vaccine antigen characterization to identify the best candidates for development. Some criteria include searching for genes that encode proteins that are surface or extracellularly secreted (for the immune system to access), are involved in the pathogenesis of the organism, are likely involved in disease-specific symptomology, are expressed during infection, and are prominent in disease-causing strains. Using a database of 1,348 complete E. coli genomes and >20,000 genomes available in public databases, we settled on the hemolysin toxin, HlyA, as a candidate worthy of exploration. Although only present in 65 genomes when blasted for the hlyA sequence (of the 1,348) (81), the phylogroup distribution of the hlyA sequence shows it is predominantly found in phylogroup B2, specifically the ExPEC and UPEC-associated sequence types (STs) ST73, ST95, ST127, and ST131 (Fig. 1A). Alignment and phylogenetic analysis of the amino acid sequences of HlyA suggests that HlyA is highly conserved, with all alleles being >97% identical on the amino acid level (Fig. 1B). These results also suggest that the hlyA sequence has a horizontal pattern of transmission between phylogroups, as the alleles of hlyA within non-B2 phylogroups are nested within those of the B2 phylogroup (Fig. 1B).

Structural prediction of HlyA

The AlphaFold2-predicted protein structure of HlyA reveals structural and organizational parallels with previously characterized RTX toxins. HlyA is predicted to comprise three domains: helix-dominated N-terminal domain (residues 1–279, red), a three-helix bundle (residues 321–437, blue), a predominantly beta-helix C-terminal domain (residues 438–1,023, green), and a linker connecting N-terminal and helix-bundle domains (residues 280–320, gray) (Fig. 1C). The electrostatic map of three-helix bundle suggests that HlyA has the capability to form membrane pores as previously seen in RTX toxins at higher concentrations; this function would be vulnerable to disruption by steric interference from antibodies that target this domain. Predominant feature of the C-terminal domain is two-strand beta helix repeats that span the length of this domain (Fig. 1D). This domain bears striking structural similarity to the beta-helical domain of highly immunogenic virulence factor pertactin from B. pertussis, which is universally used as one of the immunogenic components necessary for efficacy in acellular pertussis vaccines (107, 108). Furthermore, it has been shown that neutralizing antibodies can disrupt the interaction between RTX leukotoxin and host integrin receptors by targeting the beta helix-repeat domain mediating host-pathogen interaction [hemolysin A from P. mirabilis (top, PDB: 4W8Q) and RTX fragment from B. pertussis AC toxin (bottom, PDB: 7RAH)] (109, 110). This potentially offers an additional way to interfere with ExPEC-host interaction that could work in synergy with other vaccines with similar modes of action.

HlyA function, expression, and purification

The activation and secretion of HlyA are regulated by the hlyCABD operon, comprising the acyltransferase HlyC, the ABC transporter HlyB, and the outer membrane fusion protein HlyD (111). The secretion process can be described by the interaction of HlyA with the pre-assembled HlyBD complex, which prompts contact with TolC, a multifunctional outer membrane protein (OMP) of E. coli. This interaction forms a trans-periplasmic export channel, capable of directly transporting substrates (HlyA) from the cytoplasm to the extracellular medium, without the formation of periplasmic intermediate (112, 113). Recently, cryo-electron microscopy (cryo-EM) structures determined that the inner membrane complex formed by HlyB and HlyD is a hetero-dodecameric assembly composed of three HlyB homodimers and six HlyD subunits. Functional studies have further validated that oligomerization of HlyB and HlyD is critical for protein (HlyA) secretion (114). HlyA is a member of the RTX toxin family and possesses the ability to form a pore in the membranes of various cell types (111). However, its maturation from a non-toxic precursor, pro-HlyA, into an active toxin, necessitates a fatty acylation at two internal lysine residues (Lys-564 and Lys-690), facilitated by the acyltransferase HlyC (115). This lipidation is not required for secretion, but rather for hemolytic and cytotoxic activity. In the absence of acylation, the inactive precursor of HlyA, pro-HlyA, fails to induce pore formation in the host cell membrane (116) and does not induce calcium flux (65).

Given these reasons and with the goal of determining a safe potential vaccine candidate against ExPEC, we decided to use the non-acylated, inactive form of HlyA, pro-HlyA, as the vaccine antigen candidate, and used the co-transformation expression system (involving HlyBD and HlyA, with the sequence of hlyC deleted) to purify the protein. The plasmids we used are hlyA sequence (Uniprot entry: P08715) was cloned into plasmid pSU-hlyA, while transport complex components HlyB and HlyD were encoded into plasmid pK184-hlyBD (82). Both plasmids were co-transformed into E. coli BL21 (DE3) cells. Bacterial cultures expressing recombinant pro-HlyA antigen secreted the protein into the supernatant, which was subsequently harvested, filtered, concentrated, and visualized by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Coomassie blue staining and silver stain of the gels both revealed a dominant band at approximately 110 kDa, presumed to be pro-HlyA (highlighted in a red rectangle) (Fig. 1E). Furthermore, we included results from the control supernatant (second lane), derived from an untransformed E. coli BL21(DE3) colony cultured and purified identically to the pro-HlyA protein. Both the silver-stained and the Coomassie blue stained gels distinctly show the absence of a 110 kDa band in the control supernatant (highlighted in a red rectangle). Additionally, two extra bands were observed in the pro-HlyA lane. The first extra band (marked in a blue rectangle) was present in both the control supernatant and pro-HlyA lanes, suggesting it could be an unidentified protein from BL21 (DE3) itself, possibly degrading to around 37 kDa during the whole process. The second extra band, exclusive to the pro-HlyA lane, is hypothesized to be a degradation product of pro-HlyA (marked in a yellow rectangle) (Fig. 1E).

The identity of the putative pro-HlyA protein was confirmed by mass spectrometry by MS/MS of the band. Purified protein bands were resolved and digested in gel, and the tryptic peptides were analyzed on a nanospray LC-MS (liquid chromatography-mass spectrometry) system. The coverage of the candidate antigen pro-HlyA, which is defined as the percentage of the protein sequence covered by identified peptides, was approximately 73%, with 97 peptides detected (Fig. 1F). This high sequence coverage substantiated the identity of the antigen, enabling its utilization in further experiments. The recombinant SinH-3 protein, fused to GST, was expressed in E. coli BL21(DE3) and purified following a previously described method (83).

Pro-HlyA vaccination in a murine model of bacteremia

UTI89, an ExPEC strain belonging to multilocus sequence type 95 (ST95) (117), was isolated from a patient with acute cystitis (118). ST95, along with ST73 and ST131, is predominantly found in ExPEC strains and represents the second most prevalent clonal group in patients with BSIs (23). To evaluate the protective efficacy of pro-HlyA in a UTI89 bacteremia model, mice were subcutaneously immunized with purified pro-HlyA combined with alum adjuvant (2:1 antigen/alum ratio), while the control group mice were injected with a mixture comprising equal volumes of control supernatant and alum adjuvant. On day 42, mice were followed by intraperitoneal injection of UTI89 (1 × 10⁸ CFU/mouse). The experimental vaccination scheme is shown in Fig. 2A. After 16 hof infection, mice were euthanized, and their kidney, spleen, and liver were collected. The harvested organs were homogenized, and the bacterial burden of UTI89 in infected tissues was evaluated by quantifying CFU (Fig. 2B and C).

Combining the counts from all cohorts (to assess the total effect of vaccination across all organs), mice immunized with pro-HlyA exhibited a significant reduction in bacterial burden across all organs (Adjusted P-value, P < 0.0001) (Fig. 2B). Compared to control group mice, pro-HlyA immunized mice showed a 1.14-log reduction in median UTI89 level at 16 hours post-infection. Additionally, pro-HlyA vaccination led to significant reduction in bacterial levels within each organ type;1.65-log, 1.12-log, and 1.19-log reductions, in median UTI89 levels in the kidneys, spleen, and liver, respectively (Adjusted P-value, kidney, P = 0.0006; spleen, P = 0.0008; liver, P < 0.0001) when compared to control group mice (Fig. 2C).

Pro-HlyA immunization and lethal disease

We assessed the long-term survival rate following pro-HlyA immunization and challenge with a lethal dose of ExPEC. The experimental group mice received three subcutaneous immunizations of pro-HlyA on days 0, 14, and 28, while mice in the control group remained unvaccinated. On day 42, mice were intraperitoneally injected with UTI89 (5 × 10⁷ CFU/mouse) and monitored twice daily for morbidity and mortality over the next 10 days. At 3 days post-infection, four moribund vaccinated mice were euthanized, and organs were collected (−3 d.p.i. group) to assess changes in bacterial burden within infected organs over time. At 10 d.p.i., all remaining surviving vaccinated mice were euthanized. The harvested organs were homogenized, and UTI89 bacterial burden within infected tissues was quantified by determining CFU. The vaccination scheme used in this study is shown in Fig. 2D.

Our results indicated that unvaccinated mice died within 1 d.p.i., whereas four surviving pro-HlyA vaccinated mice were euthanized at 3 d.p.i. during the 10-day observation period (−3 d.p.i. group), and all the remaining eight Dual-Hit vaccinated mice survived until the end of the observation period (−10 d.p.i. group), resulting in a 66.7% survival rate at 10 d.p.i. (Adjusted P-value, P < 0.0001) (Fig. 2E). The bacterial burden results in pro-HlyA vaccinated mice correlated with survival rates. Combining the counts from all organs, pro-HlyA immunized mice exhibited a significant reduction in UTI89 bacterial burden at both 3 d.p.i. (Adjusted P-value, P = 0.0060) and 10 d.p.i. (Adjusted P-value, P < 0.0001), compared to unvaccinated mice. Moreover, in comparison to unvaccinated mice, the pro-HlyA vaccination resulted in approximately 5.42-log and 8.57-log reductions in median UTI89 bacterial burden at 3 d.p.i. and 10 d.p.i., respectively (Fig. 2F). Furthermore, pro-HlyA vaccinated mice surviving for 10 d.p.i. showed a 3.15-log significant reduction in median UTI89 levels (Adjusted P-value, P = 0.0147) compared to those euthanized at 3 d.p.i. Additionally, combining the counts at both 3 d.p.i. and 10 d.p.i. and analyzing by each organ, pro-HlyA immunized mice demonstrated a significant decrease in bacterial loads across multiple organs compared to unvaccinated mice (Adjusted P-value, kidney, P = 0.0013; spleen, P = 0.0040; liver, P < 0.0001) (Fig. 2G).

Efficacy of a combined autotransporter-toxin vaccine for ExPEC

We hypothesized that pro-HlyA could confer protection against strains that lack evidence for encoding the hlyA gene, potentially attributable to cross-protection induced by pro-HlyA. To evaluate this, we vaccinated the mice with pro-HlyA and conducted mortality studies, as illustrated in Fig. S1. Our results with the pro-HlyA antigen vaccine (Fig. 2; Fig. S1) demonstrate that this strategy would only provide protection against HlyA+ strains and leave subjects vulnerable to HlyA− strains. Our earlier research also demonstrates that SinH-3, a fragment corresponding to the immunoglobulin-like (Ig-like) domain-3 of the invasin-like autotransporter protein SinH, conferred robust systemic protection against infections caused by ST131 ExPEC strains in multiple murine models (83). Given the insufficient protection induced by pro-HlyA immunization alone against a mixture of five ST131 ExPEC strains that lack the hlyA sequence (including ST131-H30R lineage) (Fig. S1), we aimed to investigate the potential of a combination vaccine comprising SinH-3 and pro-HlyA (hereafter referred to as “Dual-Hit”) against several sequence types of ExPEC strains. We first assessed whether Dual-Hit still maintained robust protective efficacy against representative ExPEC strains containing hlyA sequences, such as UTI89, in both bacteremia and mortality models.

To evaluate the rapid protective efficacy of Dual-Hit immunization in a UTI89 bacteremia model, the experiment and control group mice were immunized and challenged as described in Fig. 2. The experimental vaccination scheme is shown in Fig. 3A. After 16 h of infection, mice were euthanized simultaneously, their organs were harvested and homogenized, and the bacterial burden of UTI89 was quantified by measuring CFU (Fig. 3B and C). Combining the counts from all cohorts, Dual-Hit vaccinated mice demonstrated a significant reduction in bacterial burden across all organs (Adjusted P-value, P < 0.0001) (Fig. 3B). Compared to the control group mice, Dual-Hit immunized mice exhibited an approximately 1.73-log reduction in median UTI89 level at 16 h post-infection, indicating robust and rapid protection across multiple organs. Moreover, Dual-Hit vaccination resulted in significant reductions in bacterial levels within each collected organ. Compared to control group mice, those vaccinated with Dual-Hit exhibited approximately 2.05-log, 1.72-log, and 1.61-log reductions in median UTI89 levels in the kidneys, spleen, and liver, respectively (Adjusted P-value, kidney, P = 0.0009; spleen, P = 0.0047; liver, P = 0.0289) (Fig. 3C). These findings suggest that Dual-Hit immunization can still provide rapid, systemic protection against UTI89 bacteremia across multiple organs within a short timeframe. Furthermore, the inclusion of SinH does not influence or interfere the efficacy of pro-HlyA vaccination.

Dual-Hit and lethal bacteremia

We next investigated whether subcutaneous immunization with Dual-Hit increased the survival rate of vaccinated mice when challenged with UTI89 using the same conditions as described in Fig. 2 (Fig. 3D). Our results showed that unvaccinated mice died within 1 d.p.i., whereas six moribund Dual-Hit vaccinated mice were euthanized at 3 d.p.i. during the 10-day observation period (−3 d.p.i. group), and all the remaining 12 Dual-Hit vaccinated mice survived until the end of the observation period (−10 d.p.i. group), resulting in a survival rate of 66.7% at 10 d.p.i. (adjusted P-value, P < 0.0001) (Fig. 3E). Bacterial burden measurements in Dual-Hit vaccinated mice correlated with survival rates. When combining the counts from all organs, Dual-Hit immunized mice showed a significant reduction in UTI89 bacterial burdens at both 3 d.p.i. (Adjusted P-value, P = 0.0037) and 10 d.p.i. (Adjusted P-value, P < 0.0001), compared to unvaccinated mice. Moreover, in comparison to unvaccinated mice, the median UTI89 bacterial burden in Dual-Hit vaccinated mice was approximately 4.99-log and 8.63-log lower at 3 d.p.i. and 10 d.p.i., respectively (Fig. 3F). Additionally, Dual-Hit vaccinated mice surviving for 10 d.p.i. showed a 3.65-log significant reduction in median UTI89 levels (Adjusted P-value, P = 0.0003) compared to those euthanized at 3 d.p.i. When combining bacterial burden counts at both 3 d.p.i. and 10 d.p.i. and analyzing by each organ, Dual-Hit immunized mice demonstrated significant reductions in bacterial burdens across multiple organs compared to unvaccinated mice (Adjusted P-value, kidney, P = 0.0013; spleen, P = 0.0003; liver, P < 0.0001) (Fig. 3G).

Pro-HlyA or Dual-Hit and protection against prototype strain CFT073

CFT073, a prototypical UPEC strain isolated from a female patient with acute pyelonephritis, belongs to phylogenetic group B2 and multilocus sequence type 73 (ST73) (86, 119). Notably, ST73 represents one of the most prevalent UPEC lineages, accounting for 11% and 16.6% of UPEC isolates obtained from UTI patients (including the elderly) in recent studies (120, 121). We investigated whether immunization with pro-HlyA or Dual-Hit confers robust protection against CFT073 in the murine model of mortality.

The vaccination was as described above with the scheme shown in Fig. 4A. Our results indicated that all unvaccinated mice died within 1 d.p.i., whereas mice immunized with pro-HlyA exhibited a 25% survival rate at 10 d.p.i. (adjusted P-value, P = 0.0021), while Dual-Hit vaccinated mice showed a 42% survival rate at 10 d.p.i. (adjusted P-value, P = 0.0021) (Fig. 4B). Although some vaccinated mice died within 2 d.p.i., immunization with pro-HlyA or Dual-Hit significantly improved survival rates after CFT073 infections. When combining counts from all organs, compared to unvaccinated mice, pro-HlyA immunized mice demonstrated a statistically significant reduction in CFT073 bacterial burden at 2 d.p.i. in the median level (Adjusted P-value, P = 0.0122, approximately 0.36 log reduction), suggesting certain protection caused by pro-HlyA vaccination. Additionally, pro-HlyA vaccinated mice surviving at 10 d.p.i. showed a significant reduction in the median level of CFT073 compared to those moribund or died within 2 d.p.i. (Adjusted P-value, P = 0.0023). However, there is no difference in bacterial burden between the 2 d.p.i. Dual-Hit vaccinated group mice and unvaccinated group mice (Fig. 4C). When combining bacterial burden counts at both 2 d.p.i. and 10 d.p.i. and analyzing them by each organ, pro-HlyA vaccinated mice exhibited a statistically significant reduction in bacterial burdens in the spleen (Adjusted P-value, P = 0.0146) and liver (Adjusted P-value, P = 0.0144) compared to the bacterial loads in unvaccinated mice (Fig. 4D). Similarly, Dual-Hit vaccinated mice showed a statistically significant reduction in bacterial levels in the spleen (Adjusted P-value, P = 0.0349) and liver (adjusted P-value, P = 0.0186) compared to the bacterial burden in unvaccinated mice (Fig. 4D).

Pro-HlyA or Dual-Hit and cystitis caused by UTI89 in the murine model of UTI

Urinary tract infections (UTIs) constitute a major global health concern, significantly contributing to morbidity in otherwise healthy females, with over 60% experiencing a diagnosis during their lifetime (122). In the United States, the annual incidence of physician-diagnosed UTIs exceeds 10% for females and 3% for males. UPEC is the primary causative agent, accounting for approximately 80% of UTI cases (123). Therefore, we evaluated the protective efficacy of pro-HlyA or Dual-Hit against UPEC colonization in the bladder in the murine model of UTI. Female BALB/cJ mice were immunized as previously described in Fig. 2. On day 42, mice were transurethrally inoculated with 1 × 10⁸ CFU of typical UPEC strains (UTI89 or CFT073, Fig. 5A). After 72 h of infection, bladders were harvested, homogenized, and bacterial loads of UTI89 and CFT073 were determined by quantifying CFUs.

For the experimental UPEC strain UTI89, our results demonstrated that both pro-HlyA and Dual-Hit immunizations provided robust protection against UTI89 colonization in the bladder within the UTI model. In comparison to unvaccinated mice, the pro-HlyA vaccinated mice exhibited an approximate 2.01-log reduction in the median level of UTI89 in their bladder (Adjusted P-value, P = 0.0328). Similarly, the Dual-Hit vaccinated mice showed an approximate 2.18-log reduction in the median level of UTI89 (Adjusted P-value, P = 0.0094) (Fig. 5B). However, no significant differences were observed between the experimental groups and the control group for the experimental UPEC strain CFT073 (Fig. 5C).

Dual-Hit vaccination and a mixture of 10 typical ExPEC strains

In previous studies, we demonstrated the robust protective efficacy of Dual-Hit against UTI89 (ST95) and CFT073 (ST73) in murine models of bacteremia and mortality. Additionally, we found that immunization with pro-HlyA alone provided inadequate protection against the infections caused by ExPEC ST131 strains that lack the hlyA sequence in the murine mortality model. Consequently, we evaluated whether Dual-Hit could offer robust protective efficacy and significantly increase survival rates against a mixture of ten typical ExPEC strains (including ST95, ST73, and ST131) in the murine mortality model.

To assess the protective efficacy of the Dual-Hit against multiple sequence types of ExPEC strains in the murine mortality model, the experiment group vaccinated as previously described, while a control group remained unvaccinated (Fig. 6A). At day 42, mice were intraperitoneally challenged with a composite of 10 representative ExPEC strains (1 × 10^8 CFU/mouse in total), which represent predominant sequence types with the highest virulence gene content within the ExPEC B2 phylogroup (124). These strains, present in equal proportions, span a range of common high-virulence sequence types, including CFT073 (ST73), UTI89 (ST95), W0008 (ST127), JJ1886, JJ1901, JJ2050, JJ2528, JJ2547 (ST131), W0044 (ST405-like), and W0128 (ST648-like). Notably, all used strains are SinH+, with both CFT073 and UTI89 are SinH + and HlyA+. Over the next 10 days, mice were closely monitored for morbidity and mortality twice daily.

Our findings revealed that unvaccinated mice died within 1 d.p.i., while 2 of the 18 Dual-Hit vaccinated mice died within 1 d.p.i. and 4 moribund Dual-Hit vaccinated mice were euthanized at 3 d.p.i. during the 10-day observation period (−3 d.p.i. group), and all the remaining 16 Dual-Hit vaccinated mice survived until the end of the observation period (−10 d.p.i. group), resulting in a survival rate of 72.7% at 10 d.p.i. (adjusted P-value, P < 0.0001) (Fig. 6B). When combining CFU counts from all organs, mice immunized with Dual-Hit demonstrated significantly reduced bacterial burdens at both 3 d.p.i. (Adjusted P-value, P < 0.0001) and 10 d.p.i. (Adjusted P-value, P < 0.0001) compared to unvaccinated mice. Compared to unvaccinated mice, the median bacterial burden in Dual-Hit vaccinated mice was approximately 4.07-log and 8.08-log lower at 3 d.p.i. and 10 d.p.i., respectively (Fig. 6C). A 4.01-log reduction in the median level of 10 ExPEC strains was also observed when comparing the bacterial levels in Dual-Hit vaccinated mice at 10 d.p.i. to those at 3 d.p.i. (Adjusted P-value, P = 0.0148) (Fig. 6C). Furthermore, when combining the counts at both 3 d.p.i. and 10 d.p.i. and analyzing by each organ, Dual-Hit immunized mice demonstrated a significant reduction in bacterial loads across multiple organs compared to unvaccinated mice (Adjusted P-value, kidney, P < 0.0001; spleen, P < 0.0001; liver, P < 0.0001) (Fig. 6D). Our findings indicate that incorporating SinH enhances protection against ExPEC strains lacking the hlyA sequence, suggesting that a combination vaccine will offer broader coverage of protection.

DISCUSSION

ExPEC is the predominant cause of bacteremia and UTIs, persisting in both community environments and among hospitalized patients, leading to considerable hospitalization and mortality rates. The clinical management of ExPEC faces challenges, which are further exacerbated by the overprescription of antibiotics, the emergence of antibiotic-resistant ExPEC strains, and the global aging trend (125–127). A vaccine targeting ExPEC represents a promising alternative strategy to address this issue, potentially mitigating the escalating global burden of antimicrobial resistance crisis and reducing hospitalization cost, thereby providing tremendous worldwide benefits.

In this study, we demonstrate that (i) immunization with pro-HlyA or Dual-Hit elicited a rapid and robust protection against the highly virulent ExPEC strain UTI89 (ST95), reducing the bacterial burden of UTI89 in the murine bacteremia model; (ii) immunization with pro-HlyA or Dual-Hit increased survival rates following UTI89 infection, providing lasting and consistent protection in the murine model of mortality; (iii) immunization with pro-HlyA or Dual-Hit conferred partial protection against the highly virulent ExPEC strain CFT073 (ST73), decreasing the bacterial burden of CFT073 and increasing survival rates after CFT073 infection in the murine model of mortality; (iv) immunization with pro-HlyA or Dual-Hit reduced UTI89-induced cystitis in the murine model of UTI; (v) immunization with Dual-Hit significantly increased survival rates following infection by a mixture of 10 typical ExPEC strains in the murine model of mortality, indicating the synergistic and broad-spectrum effects of the two antigens. Overall, our data indicate that both the inactive form of hemolysin, pro-HlyA, and Dual-Hit, a combination of the extracellular domains of the autotransporter SinH (SinH-3) and pro-HlyA, represent promising ExPEC vaccine candidates. We believe our findings offer an alternative research approach to the current ExPEC vaccine development efforts.

Hemolysin is a prevalent exotoxin produced by E. coli and significantly amplifies virulence in various clinical infections. Despite the relatively low abundance of hlyA in our phylogroup database as a whole, it is concentrated in highly virulent sequence types associated with ExPEC and UPEC infections, suggesting it plays a major role in these infections (Fig. 1A). Interestingly, the alignment and phylogenetic analysis of HlyA shows that the majority of instances of HlyA in what are generally considered intestine-associated phylogroups (A, B1, E) cluster together (Fig. 1B). This could suggest three different things: (i) the convergent evolution of a less-virulent (or more specialized) allele of hlyA, (ii) a more promiscuous form of the pathogenicity island carrying hlyA, and (iii) co-colonization in the intestines led to increased rates of horizontal transfer of hly. Given that the alleles found within non-B2 phylogroups are nested within the B2 phylogroup alleles, our results suggest that the B2 phylogroup acts as a reservoir for this virulence factor and that it occasionally spills over into other phylogroups, as we hypothesized previously (81).

Furthermore, while both pro-HlyA and Dual-Hit demonstrated high-efficiency protection against UTI89 in various murine models (bacteremia, mortality, and UTI), their protective efficacy against CFT073 in these models was not as robust as anticipated. The differential immunization protective efficacy against these two strains may stem from differences between UTI89 and CFT073. Although E. coli clones ST95 and ST73 frequently cause bloodstream infections and UTIs, a recent study revealed that UTI89 possesses a greater total number of genes that contribute to growth in urine and bladder colonization than CFT073. However, CFT073 appears to have more fitness factors than UTI89 (128, 129). Another study demonstrated that, while both UTI89 and CFT073 are clinical UPEC isolates that could cause infections for at least two weeks in similar proportions of mice, UTI89 infections could persist indefinitely, compared to the CFT073 infections began to clear 2 weeks after inoculation (130). The findings listed above suggest that UTI89 not only leads to more persistent infections but also might increase the detectability and binding affinity for vaccine-specific antibodies targeting UTI89, consequently enhancing the protective efficacy of pro-HlyA and Dual-Hit immunization. Moreover, variations in lipopolysaccharide (LPS) composition across bacterial strains also affect bacterial characteristics and interactions with host organisms, resulting in varied innate and adaptive immune responses (131, 132). These variations may also account for the observed differences in the protective efficacy of our antigenic formulations between the UTI89 and CFT073 strains. We also recognize that a limitation of our study is that kidney bacterial burdens were not assessed, thus future investigations of the UTI model warrant simultaneous collection of both bladder and kidney tissues. Additionally, we also evaluated the protective efficacy of both antigen forms in diminishing colonization by UTI89 or CFT073 strains in the gastrointestinal (GI) tract model. Unfortunately, neither antigen form provided protection against the ExPEC strains when compared to the control group. These results necessitate a focus on optimizing immunization routes and adjuvants to enhance mucosal immunity as a priority for future research (Fig. S3).

Variations in immunization routes and adjuvant types significantly impact vaccine efficacy evaluation. Understanding these would enable optimization of the vaccine formulation and administration to maximize protective efficacy. For instance, intramuscular (I.M.) administration is the most used route for licensed vaccines and has been shown to elicit high immunogenicity in adult rabbits immunized with MecVax, producing antibodies against enterotoxigenic Escherichia coli (ETEC) H10407 and reducing intestinal colonization (133). A clinical trial also demonstrated the safety and immunogenicity of the CS6-targeted candidate vaccine, CssBA, when administered intramuscularly (134). Furthermore, both experimental and clinical evidence have indicated that mucosal immunization could efficiently induce local and distant systemic immune responses, as well as in the blood (135, 136). In previous studies, mice intranasally immunized with the iron receptor, FyuA, IutA, Hma, and IreA exhibited a robust and long-lived humoral immune response against UPEC challenge. Intranasal immunization with FyuA reduced UPEC strain 536 colonization following transurethral challenge, while IreA intranasally immunization significantly reduced CFT073 bacterial counts in the bladder (45, 46). Therefore, without compromising the robust systemic protection provided by pro-HlyA or Dual-Hit immunization, combining subcutaneous with either intramuscular or intranasal routes may present a more promising approach to enhance immune responses and improve protective efficacy of vaccinated mice against ExPEC infections in both blood (bacteremia) and mucosal (urinary tract).

The potential of a vaccine could also be significantly enhanced by formulating it with novel adjuvants, which effectively augment immune responses to the administered vaccine antigen (137). In our study, we utilized aluminum salts (alum) as the adjuvant due to its proven safety. Alum is a clinically approved and widely used adjuvant in human vaccines, has been used for over 80 years in vaccine research, and typically stimulates the Th2-type immune responses (138). In a previous study, suitable adjuvants were screened for iron receptor-based immunization against UPEC infection, and they found that dmLT generated the most consistently robust antibody response in intranasally immunized mice, while Monophosphoryl-Lipid A (MPLA) and alum produced greater concentrations of antigen-specific IgG with intramuscular immunization (48). This study suggests that dmLT, a mucosal adjuvant proven safe and potent through both preclinical and clinical studies, could be an alternative adjuvant in future research. Similarity, a recent study demonstrated that following bladder infection, highly T-helper type 2 (TH2)-skewed immune responses prioritized bladder epithelial repair after extensive exfoliation of epithelial cells, rather than bacterial clearance and even inhibition of Th1-mediated responses (139). Therefore, MPLA is a potentially ideal adjuvant that could safely induce an appropriate level of Th1 response, enhancing Th1-mediated bacteria-clearing responses and increasing the ability to eliminate E. coli infection after vaccine immunization (140). It will be of great value to determine whether alternative adjuvants improve upon the results we have seen herein with alum, as well as if different routes of vaccination change outcomes. Of note, we did generate a mRNA version of the Dual-Hit vaccine reported here and tested it for efficacy against UTI89. However, there was no difference in survival or protection between vaccinated and control groups (Fig. S2).

In summary, antimicrobial resistance is a leading threat to global health. Developing an ExPEC vaccine presents a promising strategy to combat this growing global crisis and effectively reduce the incidence of antibiotic-resistant ExPEC infections to improve public health outcomes. Our study demonstrates the promising protective efficacy of pro-HlyA and Dual-Hit immunizations against ExPEC infections in murine models. Furthermore, in this study, we bridged computational genomics and virulome vaccinology and attempted to block various steps of bacterial pathogenesis by synergizing multiple protein subunits associated with different virulence factors. The observed reduction in bacterial burden and increased survival rates indicate the potential of these vaccine candidates for further development and evaluation.

ETHICS APPROVAL

All methods performed on mice adhered to the relevant guidelines and regulations set forth by “The Guide and Care and Use of Laboratory Animals” (National Institute of Health). The Animal Use Protocol number AN-5177 was approved by Baylor College of Medicine’s Institutional Animal Care and Use Committee.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/iai.00440-23.

Fig. S1. iai.00440-23-s0001.docx.

Evaluating the protective efficacy of pro-HlyA antigen alone against the mixture of ST131 ExPEC strains infection in the murine model of mortality.

Fig. S2. iai.00440-23-s0002.docx.

Evaluating the protective efficacy of Dual-Hit mRNA vaccine against UTI89 infection in the murine model of mortality.

Fig. S3. iai.00440-23-s0003.docx.

Assessment of the protective efficacy of pro-HlyA or Dual-Hit against either UTI89 or CFT073 colonization in the murine model of gastrointestinal (GI) tract.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.