FimH, CsgA, and PapG protein sequences from the E. coli strain CFT073 were obtained from the NCBI database (http://www.ncbi.nlm.nih.gov/protein) under accession numbers AAN83822.1, AAN79779.1, and AAN82031.1, respectively. Signal peptide prediction was performed for each protein using the SignalP 4.1 server, and five repetitions of the EAAAK sequence were employed to fuse the FimH, CsgA, and PapG proteins based on the linker library construction proposed by Li et al. (2016).

In addition, the molecular weight, theoretical isoelectric point (pI), amino acid composition, estimated lifetime, aliphatic index, and grand average of hydropathicity (GRAVY) of the fusion proteins were determined using the ProtParam program of ExPASy (Wilkins et al., 1999). The Codon Adaptation Index (CAI) and guanine and cytosine (GC) content of the genes were determined using the OPTIMIZER program (Puigbò et al., 2007). The secondary structure of the fusion proteins was predicted using the GOR IV program (Sen et al., 2005). Fusion protein modeling was performed using the hierarchical modeling approach (I-TASSER) and visualized with PyMOL software (Yang et al., 2015). Tridimensional (3D) structures were refined and minimized with KoBaMIN (http://csb.stanford.edu/kobamin/) and VegaZZ (NAMD; Pedretti et al., 2004). The 3D models were then validated by Protein Structure Analysis (ProSA) to determine Z-scores and Ramachandran plots using PROCHECK (Laskowski et al., 1998; Wiederstein and Sippl, 2007). Merging of the 3D fusion proteins with the crystal structures of the mannose-binding domain FimH (Protein Data Bank; 1TR7) and lectin domain PapG (PDB; 1J8R) were calculated with the root mean square deviation (RMSD) using the TM-align program (Zhang and Skolnick, 2005).

Immune response was theoretically determined to establish variants of fusion proteins with enhanced ability to generate an immune response. The primary and secondary structures of the fusion proteins were employed to determine the lineal antigenic epitopes of B with the AbcPred server (http://www.imtech.res.in/raghava/abcpred/), and peptides with affinities for major histocompatibility complex (MHC) class II were identified with the NetMHCII program (http://www.cbs.dtu.dk/services/NetMHCII/). The 3D structures of the fusion proteins were employed to determine the conformational antigenic epitopes using the Discotope server (http://www.cbs.dtu.dk/services/DiscoTope/).

The gene sequences of fimH, csgA, and papG from the E. coli strain CFT073 were obtained from GenBank through NCBI (http://www.ncbi.nlm.nih.gov/genbank/) under the accession numbers GQ487191.1, NC_004431.1, and AF447814.1, respectively. The conserved sequences of the fimH, csgA, and papG genes from various UPEC strains (UTI89, ABU83972, NA114, UPEC 26-1, CF-088, CF-468, isolates IA2, and AD110) were determined using the BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and Clustal Omega (http://www.ebi.ac.uk/Tools/msa/clustalo/) programs. Consensus nucleotide sequences were fused with the appropriate codons from the sequence EAAAK to generate a template fusion gene. The sequence of the template fusion gene was optimized using the OPTIMIZER program (http://genomes.urv.es/OPTIMIZER/), and messenger RNA secondary structures of monomeric, dimeric, and trimeric genes were predicted using the Mfold program (http://unafold.rna.albany.edu/?q=mfold/RNA-Folding-Form). The optimized template fusion gene, containing a 5′ BamHI site and 3′ SacI site, was synthesized by GenScript (Piscataway, NJ, USA) and cloned into the pUC57 vector to amplify and clone monomeric, dimeric, and trimeric genes.

Primer designs for monomeric, dimeric, and trimeric genes from the synthetized template fusion gene were generated manually according to the aLICator LIC Cloning and Expression handbook (Thermo Fisher Scientific, Waltham, MA, USA) and synthesized by IDT Technologies (Coralville, Iowa, USA). Gene amplification was performed by polymerase chain reaction (PCR) using Pfu DNA polymerase. Conditions for amplification and cloning were obtained from the aLICator LIC Cloning and Expression handbook. The pLATE31 expression vector was used to clone the fusion genes and transformed by electroporation into BL21 (DE3) E. coli.

Fusion genes in recombinant vectors were verified by next-generation sequencing on a NexSeq 500 system (Illumina, San Diego, CA, USA). Specific sequencing primers were obtained from the aLICator LIC Cloning and Expression Kit, and the gene sequences were analyzed by BLAST and submitted to GenBank at NCBI.

A BL21 E. coli strain carrying the pLATE31 expression vector (fusion proteins) was plated on Luria Bertani (LB; Becton-Dickinson, Franklin Lakes, New Jersey, USA) agar and incubated for 16 h at 30°C. The putative transformant colonies were selected via the colony-blotting method using an anti-6X His (C-Term) HRP antibody (Abcam; Cambridge, MA, USA). Solubility assays were performed following the protocol described in the QIAexpressionist manual from Qiagen (Jacques-Schiesser-Str, Stockach, Germany). Fusion protein expression was performed in 500 mL of LB medium supplemented with 1 mM IPTG for 5 h at 37°C. The cell pellets were resuspended in phosphate buffer (10 mM K₂HPO₄, pH 7.4, 150 mM NaCl, and 1 mM EDTA) with phenylmethylsulfonyl fluoride (PMSF; Sigma-Aldrich Corp., St. Louis, MO, USA), lysed by sonication, and centrifuged at 26,116 g for 20 min. The supernatants were discarded, and the pellets were resuspended in denaturing buffer (8 M guanidine, 100 mM NaCl, and 100 mM K₂HPO₄, pH 8). After 3 days of incubation at room temperature, the lysates were centrifuged at 58,762 g for 20 min. The supernatants were incubated on a column containing nickel-nitrilotriacetic acid-agarose (Qiagen) at 4°C for 1 h, washed with A-buffer (8.5 M urea, 20 mM Tris, pH 7.5, 160 mM NaCl, and 20 mM imidazole) and eluted with B-buffer (8 M urea, 50 mM Na₂HPO₄, pH 8, 100 mM NaCl, and 500 mM imidazole). The collected proteins underwent refolding via dialysis using a urea gradient from 7 to 1 M in C-buffer (25 mM Tris, pH 7.5, 100 mM NaCl, and 0.5 mM EDTA); incubation was performed at 4°C for 24 h. Refolded fusion proteins were stored in C-buffer at −70°C for all assays.

Fusion proteins were quantified according to the protocol for the 2D-Quant kit (GE Healthcare Bio-Sciences AB, Björkgatan, Uppsala, Sweden), separated by performing 14% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), visualized by Coomassie staining and identified by mass spectroscopy using a 4800 MALDI TOF/TOF™ Analyzer (Applied Biosystems/MDS SCIEX, Waltham, MA, USA) after excision of the proteins spots. The CsgA protein was treated with 88% formic acid (Sigma-Aldrich Corp., St. Louis, MO, USA) prior to SDS-PAGE (Saldaña et al., 2009). Molecular weights were estimated using Image Lab software version 5.2 from Bio-Rad (Hercules, California, USA). The aggregation state of fusion proteins was determined by performing dynamic light scattering (DLS) using a Zetasizer Helix (Malvern Instruments Ltd, Grovewood Road, Worcestershire, United Kingdom). Fusion proteins with histidine tags were transferred onto polyvinylidene difluoride (PVDF) membranes and confirmed by performing Western blot assays using anti-6His (C-Terminal) HRP antibodies (Abcam; Cambridge, MA, USA) as described by Ledesma et al. (2010). The endotoxin (LPS) levels of the purified fusion proteins were determined using the Pierce™ LAL Chromogenic Endotoxin Quantitation kit according to the manufacturer's protocol (Thermo Fisher Scientific; Waltham, MA, USA). Additionally, fusion proteins were treated with 50 μg/mL polymyxin B (Sigma-Aldrich Corp., St. Louis, MO, USA) for 12 h at 4°C prior to bioactivity assays.

TLR2 and TLR4 expression in HTB5 bladder cells [American Type Culture Collection (ATCC), Manassas, VA, USA] was analyzed by flow cytometry using a human TLR2 fluorescein-conjugated antibody (R&D Systems, Inc., Minneapolis, USA) and a human TLR4/MD-2 complex phycoerythrin-conjugated antibody (Santa Cruz Biotechnology Inc., Texas, USA). Activation of TLR2 and TLR4 by the fusion proteins was assessed by quantifying the release of IL-6 and IL-8. HTB5 cells were cultured in 24 well-plates (Greiner, Germany) at a density of 10⁵ cells/well incubated with 1 mL of fresh Eagle's minimum essential medium (EMEM; ATCC® 30-2003™) supplemented with 10% fetal bovine serum (FBS) from Gibco (Thermo Fisher Scientific; Waltham, MA, USA). Cytokine induction in HTB5 cells was detected after 6 h of incubation with 10 μg/mL FimH, CsgA, PapG, FC, and FCP proteins by performing enzyme-linked immunosorbent assays (ELISAs) following the protocol established by BD Biosciences (San Jose, CA, USA). In addition, 100 ng/mL lipopolysaccharide (LPS; Sigma-Aldrich Corp., St. Louis, MO, USA) from E. coli 0111:B4 and 100 ng/mL lipoteichoic acid (LTA; Sigma-Aldrich Corp., St. Louis, MO, USA) from S. aureus were used as controls for the induction of TLR4 and TLR2, respectively.

The study was approved by the Research (Dr. Onofre Muñoz Hernández), Ethics (Dr. Amparo Faure Fontenla), and Biosecurity (Dr. Herlinda Vera Hermosillo) Committees of Hospital Infantil de México Federico Gómez (HIMFG) under numbers HIM/2014/022 and HIM/2016/027. Physicians from the Infectology Department obtained permission and consent from patients to use the urine and serum samples employed in this study. Serum and urine samples were collected from 14 UTI patients (UP); samples from 14 healthy patients (HP) recruited from “Laboratorio Central del HIMFG” were used as negative controls. Patients were selected based on the following criteria: UTI symptoms, urine culture with ≥100,000 CFU/mL E. coli, leukocyte esterase and/or nitrites, and urothelial cells in urine. Urine samples were centrifuged at 7835 g for 5 min and filtered through a 0.22-μm Durapore membrane (Merck Millipore; Darmstadt, Germany). IgG and IgA antibody titers against the FimH, CsgA, PapG, FC, and FCP proteins were determined by ELISA using serum and urine samples diluted 1:50 and 1:10, respectively.

Five female 6-month-old New Zealand rabbits were obtained from the “Instituto de Fisiología Celular de la Universidad Nacional Autónoma de México” for use in this study. Initially, the rabbits were immunized subcutaneously with 200 μg of FimH, CsgA, PapG, FC, and FCP proteins in complete Freund's adjuvant. The rabbits were re-immunized three times (days 21, 28, and 37) with 100 μg of each protein in incomplete Freund's adjuvant and bled via cardiac puncture on day 40. The collected blood was centrifuged at 7835 g for 5 min, and separated sera were stored at −70°C until use. The sera were absorbed using the CFT073 csgA::km + fimH::cm mutant strain generated in this study via a one-step inactivation method (Datsenko and Wanner, 2000). Anti-PapG and anti-FCP polyclonal rabbit sera were adsorbed using the CFT073 csgA::km + fimH::cm mutant strain under conditions that did not permit PapG expression, as confirmed by RT-PCR. Additionally, sera were heat-inactivated at 56°C for 30 min and titrated by ELISA using serial dilutions from 1:10 to 1:100,000 against the appropriate proteins.

HTB5 bladder cells were cultured in 24 well-plates at a density of 10⁵ cells/well in 1 mL of fresh EMEM supplemented with 10% FBS until 80% confluence was reached. Previously, the strain CFT073 was grown overnight in LB broth at 37°C, and an aliquot (1:100 dilution) of the culture was incubated until it reached an OD₆₀₀ of 1.0. The bacterial culture was then mixed with pre-immune or/and immune serum at a 1:1 ratio and maintained 2 h at 37°C with constant agitation. Cell monolayers were infected with 10⁷ bacterial cells [multiplicity of infection (MOI): 100] and incubated for 2 h at 37°C in 5% CO₂. Infected cell monolayers were washed three times with PBS, and 200 μL of PBS containing 0.1% Triton X-100 was added. The bacteria that adhered to the cell monolayers were homogenized, and colony-forming units were determined using the Spotting Method described by Hannan and Hunstad (2016).

One-way ANOVA and Student's t-test were used to compare differences between the mean and median values of groups using GraphPad Prism software (version 6). A p < 0.05 was considered statistically significant for all results.

Fusion proteins were generated using the sequences of the fimH, papG, and csgA genes, which exhibit identity of ≥90% among clinical UPEC strains. In addition, a signal peptide was identified between residues 1 and 27 of the sequences of the FimH and PapG proteins and residues 1–21 of the CsgA protein. Additionally, the CsgA protein contained a nucleation region between residues 22 and 41. These signal peptides and the nucleation region were eliminated from the sequences used to design and synthesize of the fusion proteins. The fusion of the FimH, PapG, and CsgA proteins was generated by employing the five repeats of the EAAAK linker ([EAAAK]₅), which forms an alpha helix with a length of 39.95 Ȧ stabilized by five salt bridges between the Glu⁻ and Lys⁺ residues.

A template fusion gene containing (in order) fimH-csgA-papG-fimH-csgA (fcpfc) linked to the nucleotide sequence of the [EAAAK]₅ peptide was designed (Figure 1A). Six specific primers localized in the 5′ (forward) and 3′ (reverse) regions of each gene were designed and synthesized as described in the aLICator LIC Cloning and Expression Handbook (Table 1). Monomeric, dimeric, and trimeric genes were generated using combinations of the six primers in PCR amplification and cloned into the pLATE31 expression plasmid (Figure 1B).

Increased expression levels of the fusion genes were obtained by preferential codon optimization of the fcpfc template fusion gene, which exhibited a CAI of 0.433–1.0 and a GC content of 48.9–55.4%. The fcpfc-optimized gene was synthesized chemically and cloned into the pUC57 vector to produce pF₂C₂P. Sequencing and digestion of the pF₂C₂P were consistent with the theoretical profile (data not shown).

The physical and chemical parameters of the monomeric (FimH, CsgA, and PapG), dimeric [FimH-CsgA (FC), CsgA-PapG (CP), and PapG-FimH (PF)], and trimeric [FimH-PapG-CsgA (FCP), PapG-CsgA-FimH (PCF), and CsgA-FimH-PapG (CFP)] proteins are shown in Table 2. Primary sequence analysis of these proteins revealed a high number of linear epitopes, between 18 and 33, as well as a high number of peptides binding to MHC-II, between 68 and 159 (Table 3). The secondary structure predictions for the monomeric, dimeric, and trimeric proteins indicated a high percentage of random coil and beta-sheet structures; however, lower percentages of alpha helices were observed (Figure 2). The conserved region of the alpha helix structure conferred by the [EAAAK]₅ linker was identified in all fusion proteins.

The 3D structure models of the dimeric and trimeric proteins were refined 10 times to obtained the optimal minimization energy (Figure 2 and Table 4). Interestingly, the FC and FCP proteins exhibited models with optimal theoretical data that indicated a stable structural conformation. Gibbs free energy (ΔG) data indicated high stability for the FC protein, with ΔG = −7040.78 kcal/mol, and the FCP protein, with ΔG = −13,343.83 kcal/mol (Table 4). The 3D model validation of the FC and FCP proteins indicated that 94.7 and 93.1% of the amino acid residues were within permitted regions in the Ramachandran plot, with Z-scores of −6.95 (FC) and −9.53 (FCP).

The 3D models of the FC and FCP proteins exhibited RMSDs of 1.81 and 1.67 Å, respectively, indicating conformations very similar to that of the mannose-binding domain of FimH. The FCP protein also exhibited an RMSD of 1.62 Å with the lectin domain of PapG (Table 4). Analysis of the 3D models using the Discotope program revealed 13 and 17 conformational epitopes in the FC and FCP proteins, respectively (Table 3).

Bioinformatic analysis was an essential tool for the selection of monomeric (fimH, csgA, and papG), dimeric (fc), and trimeric (fcp) genes, which were cloned in the pLATE31 expression vector. PCR assays were performed to verify the cloned expression vectors using specific sequencing primers as described in the aLICator LIC Cloning and Expression Handbook (Figure 3A). The sequences of the fusion genes were determined by sequencing and will be submitted to the GenBank database (unpublished data). The BL21 (DE3) E. coli strain carrying the expression vectors was induced with 1 mM IPTG, and the His-tagged fusion proteins were purified under denaturing conditions by affinity chromatography. Coomassie blue-stained SDS-PAGE gels and Western blot analysis revealed 26.5, 11.9, 33.9, 44.9, and 82.1 kDa bands corresponding to FimH, CsgA, PapG, FC, and FCP, respectively (Figures 3B,C). The protein concentrations of FimH, CsgA, PapG, FC, and FCP were 2.7, 0.697, 2.63, 1.03, 0.998 mg/mL, respectively.

Mass spectrometry analysis by MALDI-TOF/TOF revealed the presence of specific peptides corresponding to the fusion proteins, and DLS analysis of the fusion proteins indicated a polydispersed state (Table 5).

Flow cytometry analysis of HTB5 bladder cells revealed an expression of 50.3% for TLR2 and 31.7% for TLR4 (Figure 4). Endotoxin levels ≤ 0.012 EU/mL were observed when the FimH, CsgA, PapG, FC, and FCP proteins were treated with 50 μg/mL polymyxin B. HTB5 bladder cells expressing TLR2 and TLR4 were incubated with 10 μg/mL FimH, CsgA, PapG, FC, or FCP, which induced different percentages of IL-6 and IL-8 cytokine release. Briefly, the FimH, CsgA, and PapG proteins stimulated the release of 372–398 pg/mL IL-6, and no significant differences in IL-6 release were observed among these proteins. However, the FC and FCP proteins induced the release of 464.79 and 521.24 pg/mL IL-6, respectively, which were significantly different (p ≤ 0.018 and p ≤ 0.002, respectively) from the values obtained for the monomeric proteins (Figure 5). In addition, 398.52 pg/mL IL-8 was released by the FC protein (p ≤ 0.003 compared to the monomeric protein), and 450.40 pg/mL IL-8 was released by the FCP protein (p ≤ 0.037 compared to the monomeric protein; Figure 6).

ELISA assays were performed to detect antibodies against the FimH, CsgA, PapG, FC, and FCP proteins in the sera and urine of HP and UP (Figures 7, 8). Antibodies in the sera reacted with all proteins evaluated. High levels of UP-IgA antibodies in sera were observed, with a median OD₄₅₀ of 1.628–2.216. Interestingly, high levels of UP-IgA antibodies in sera, with a median OD₄₅₀ of 2.216 were detected when ELISA was performed using the FC protein as an antigen (Figure 7B). However, under identical conditions, low levels of UP-IgG (median values of 0.514–0.837) were detected. The FC protein exhibited greater specificity, with sera detecting UP-IgG with a median value of 0.837 (Figure 7A). Additionally, all fusion proteins reacted with high levels of UP-IgA antibodies in urine, with a significant difference of p = 0.0003 compared to the values for HP-IgA antibodies. Furthermore, 1 μg/mL fusion protein detected high levels of UP-IgA antibodies in urine, with median values ranging from 0.6189 to 0.950 (Figure 8B). In contrast, low levels of UP-IgG were detected in urine, with median values ranging from 0.444 to 0.619 (Figure 8A). The difference in the values between UP-IgA and UP-IgG in urine was significant (p < 0.05) compared to HP-IgA and HP-IgG (Figure 8).

Polyclonal rabbit antibodies against the FimH, CsgA, PapG, FC, and FCP proteins were used to block bacterial adherence to HTB5 bladder cells. The strain CFT073 showed adherence to 6.3 × 10⁶ HTB5 bladder cells, and this level of adherence was set at 100%. Similar adherence by the strain CFT073 was observed with pooled pre-immune sera (Figure 9). Bacterial adherence in the presence of polyclonal rabbit antibodies showed a significant reduction to 32% for anti-FimH (p = 0.002), 21% for anti-CsgA (p = 0.0011), 60% for anti-PapG (p < 0.0001), 73% for anti-FC (p < 0.0001), and 46% for anti-FCP (p < 0.0001) compared with the CFT073 and pre-immune serum controls (Figure 9).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.