First, immunoinformatic analysis was implemented for systematic identification of potential epitopes and antigens for tularemia vaccine development. The genome information of 756 strains of Francisella with a normal genome size and number of protein-encoding sequences (CDS) was obtained from the NCBI database. The mean genome size was 1.80 Mb (1.78-1.9 Mb), and the average number of CDS was 1,519 ( Figure 1A ). Then, the Clusters of Orthologous Groups (COG) functional annotation of 2,188 coding sequences of the representative highly virulent Ft subsp. tularensis Schu S4 strain was applied, and 1,703 identified proteins had known functional classifications, covering 22 functional classifications in COG. Among them, the clusters for “translation, ribosomal structure and biogenesis” (184, 10.8%), “replication, recombination and repair” (158, 9.2%) and “amino acid transport and metabolism” (153, 9.0%) were the top three largest functional groups ( Figure 1B ). Then, the 95 proteins that carried one or more protective signatures recurring in known bacterial protective antigens (Altindis et al., 2015) were further analyzed using two reverse vaccinology tools, including the self-developed Multifactor Prediction of Protective Antigens (MPPA) platform based on subcellular localization, antigen similarity, antigenicity, mature epitope density, virulence, and adhesion probability (Zai et al., 2021) and the Vax-ELAN pipeline based on subcellular localization, transmembrane helix prediction, adhesion property, non-homology with host proteins, etc (Rawal et al., 2021) ( Figure 1C ). The potential immunogenic antigens—Tul4, OmpA, FopA, and DnaK—had both a high MPPA score and Vax-ELAN value, indicating likely involvement in antigenicity and virulence, and were further assessed as vaccine candidates.

Codon-optimized Tul4, FopA, and DnaK were purified and examined for immunogenicity when formulated as subunit vaccines ( Figure 2A ). Mice that were subcutaneously (s.c.) immunized three times 2 weeks apart with 10 μg of recombinant Tul4, FopA, or DnaK together with Al(OH)₃+CPG adjuvants had high self-matched IgG antibody responses, while cocktail (Tul4+FopA+DnaK) immunization showed no superiority ( Figures 2B, C ). The IgG levels in vaccinated mice rose steadily, while the IgM levels peaked 2 weeks after the first immunization and subsequently declined in all experimental groups ( Figure 2C ). Moreover, indistinguishable increased anti-iFt IgG antibody responses were observed with increasing time among the different groups ( Figure 2D ). However, all immunized mice succumbed to infection by 6-7 days after challenge, almost without exception ( Figure 2E ). In agreement with previous studies (Kaur et al., 2012), these results demonstrated that recombinant Tul4, FopA, and DnaK subunit vaccines conferred immunogenic potential but little protection against challenge with high-activity Ft LVS.

Then, the recombinant Ad5-Tul4, Ad5-OmpA, Ad5-FopA, and Ad5-DnaK vaccines were further analyzed for immune reactivity. Mice receiving a single dose of Ad5-Tul4, Ad5-OmpA, Ad5-FopA, or Ad5-DnaK by the i.m. route generated increased anti-iFt IgG levels with increasing time ( Figures 3A, B ). Coadministration of a mixture of various Ad5-based vaccines also resulted in no significant augmentation of the serum IgG immune response ( Figure 3B ). Furthermore, Ad5-based vaccines generated a Th1-predominated humoral immune response, indicated by higher levels of Ft-specific IgG2a antibodies ( Figure 3C ). In addition, splenocytes from mice immunized with Ad5-Tul4 (14 days post immunization) responded to in vitro stimulation with purified Tul4 fragment 17 (QGSVRLQWQAPEGSK) and to a greater extent fragment 18 (RLQWQAPEGSKCHDT) by producing IFN-γ, demonstrating that the Tul4 antigen was indeed processed and presented to T cells, eliciting a potent cellular immune response ( Supplementary Figures S1A, B ). Remarkably, 50% of the mice after single immunization with Ad5-Tul4 or Ad5-mixture were protected against Ft LVS challenge, while a relatively lower survival rate (25%) was observed when mice were immunized with Ad5-OmpA, Ad5-FopA, or Ad5-DnaK ( Figure 3D ). Taken together, these results demonstrated the ability of Ad5-based vaccines, especially Ad5-Tul4, to provide efficient protection against Ft LVS.

To determine the optimal immunization regimen for protection, groups of mice were immunized via the i.n. route and i.m. route with Ad5-Tul4 on days 0 and 21 ( Figure 4A ). Two intranasal immunizations induced IL-17A production, while intramuscular immunizations tended to elicit a higher level of IL-2 production, indicating that Th17- and Th1-predominated humoral immune responses were induced by i.n. and i.m. immunization, respectively ( Figure 4B ). Mice immunized with Ad5-Tul4 via the i.n. route produced higher levels of anti-iFt IgG and IgA in bronchoalveolar lavage fluid, highlighting that intranasal delivery induced a remarkable localized mucosal immune response ( Figure 4C ). Immunized and nonimmunized mice were then challenged via the respiratory route or intraperitoneal route to mimic mucosal and systemic infection, respectively, with Ft LVS (2×10³ CFU) 3 weeks after the second immunization. Five days post-infection, the spleen, liver and lung were harvested, and the expression of Ft LVS-specific AKR DNA was assessed using real-time PCR as a measurement of bacterial burden in the tissues. Compared to the i.m. immunized mice, i.n.-immunized mice showed a significantly lower bacterial dissemination in the lung and spleen upon i.n. challenge, with a 73.86% Ft LVS load reduction in the lung, 92.81% in the spleen and 95.16% in the liver ( Figure 4D , Supplementary Table S1 ), demonstrating improved protection with respiratory vaccination against intranasal challenge. However, the effect on of i.n. immunization on bacterial clearance was poor upon i.p. challenge, the Ft LVS load reduction rates were 52.48%, 45%, and 17.38% in the lung, spleen, and liver, respectively, much lower than that with i.m. immunization (Ft LVS load reduction rates were 99.31%, 99.70%, and 99.81%, respectively) ( Figure 4D , Supplementary Table S1 ). Consistently, approximately 80% protection against i.n. challenge was provided by both i.n. and i.m. immunization, while only i.m.-immunized mice survived i.p. infection ( Figure 4E ). Taken together, these results demonstrated that a regimen consisting of two intranasal immunizations with Ad5-Tul4 was more effective in protecting mice from respiratory infection, while intramuscular immunizations were superior for resisting systemic Ft colonization.

The Francisella tularensis LVS strain was obtained from the Institute of Microbiology and Epidemiology, Beijing, China (Duan et al., 2000), handled in a BSL-2 laboratory; cultured on solid medium containing 4% tryptic soy agar (TSA, Solarbio, T8650), 0.2% L-cysteine hydrochloride monohydrate (Solarbio, C0011), and 7% defibrinated rabbit blood (Shanghai Yuanye Bio-Technology Co Ltd, MP20025); and grown for 12-15 h at 37°C in an atmosphere of 5% CO₂. Then, a bacterial suspension was made in sterile PBS, followed by OD₆₀₀ measurement (one OD₆₀₀ corresponded to 3× 10⁸ CFU/ml). The bacteria were diluted to achieve a final concentration of 2× 10⁴ CFU/ml or 4× 10⁴ CFU/ml, and these suspensions were used in all challenge experiments.

pTIG-Trx vectors were constructed with the codon-optimized Tul4, FopA, and DnaK genes. Then, the vectors were transferred to E. coli BL21 competent cells, and expression was induced by the addition of IPTG (isopropyl-β-d-thiogalactoside) when the OD₆₀₀ reached 0.8-1.0. The identity of the purified rTul4, rFopA, and rDnaK proteins was examined by sodium dodecyl sulfate−polyacrylamide gel electrophoresis (SDS−PAGE) with Coomassie Blue staining. The concentration of each recombinant protein was estimated by a BCA Protein Assay Kit (Thermo Fisher, USA) according to the manufacturer’s instructions.

Recombinant replication-defective human type 5 adenoviruses were constructed as described previously (Wu et al., 2020). Briefly, codon-optimized genes encoding Tul4, OmpA, FopA, and DnaK were synthesized and cloned into the shuttle plasmid pDC316 with the AdMax adenovirus system (Microbix Biosystem, Canada). HEK293 cells were then cotransfected with the constructed shuttle plasmids together with the backbone plasmid (pBHGloxΔE1, 3Cre). The cells exhibiting obvious cytopathic effects were lysed, and the obtained recombinant adenoviruses were amplified by serial passage in HEK293 cells. The viruses were purified by an Adeno-X™ Virus Purification Kit (BD Biosciences, Clontech) and titrated by an endpoint dilution assay.

Animal experiments were performed according to the guidelines of the Institutional Experimental Animal Welfare and Ethics Committee. Female BALB/c mice aged 8-10 weeks (purchased from Vital River Laboratories, Beijing, China) were used for all experimental groups. For subunit vaccine administration, mice were immunized via the s.c. route on days 0, 14 and 28 with 10 μg of the indicated proteins together with 100 μg of Al(OH)₃ (Alhydrogel^®2%, Brenntag Biosector, Frederikssund, Denmark) and 25 μg of CpG1826 (5’- TCCATGACGTTCCTGACGTT-3’, Takara Clontech) adjuvants. For Ad5-based vaccine administration, mice were immunized i.m. (100 μl) or i.n. (50 μl, 25 μl per nostril) on day 0 or on days 0 and 21 with 1x10⁸ plaque-forming units (PFU) of the indicated adenoviruses. For challenges, intraperitoneal (i.p.) injection (100 μl, 2×10⁴ CFU/ml) or intranasal inhalation (50 μl, 25 μl per nostril, 4×10⁴ CFU/ml) of Ft LVS were applied at the indicated time.

Ft burden was evaluated with real-time PCR as previously described (Shi et al., 2009). Briefly, total nucleic acids were purified from the lung, spleen or liver harvested from euthanized mice by phenol−chloroform extraction and then quantified by a NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, DE), followed by dilution to 5 ng/μl. The Ft-specific AKR gene was quantified by using a TaqMan Universal Master Mix II Kit (Thermo Fisher, USA). Five microliters of purified DNA was amplified in a 20 μl reaction containing 10 μl of 2×TaqMan Universal Master Mix II, 0.8 μM forward primer, 0.8 μM reverse primer, and 200 nM probe. A forward primer (5’-GCAGGGCGAGCACCATT-3′), reverse primer (5′-ATCTTGCATGGTCACCACTTGA-3’), and probe (5’-FAM-CGATATTTGCCTGTTAGCACTCCT-Tamra-3’) were used. Reactions were incubated at 50°C for 2 min, followed by 95°C for 10 min, and then thermal cycled for 40 cycles (95°C for 15 s and 60°C for 1 min).

For measurement of IgG (IgG1 and IgG2a) and IgM titers, serum from immunized mice was serially diluted and added to 96-well microplates (Corning, USA), which were precoated with 2 μg/ml indicated antigen proteins or 4% paraformaldehyde-inactivated Ft (iFt) at 4°C overnight. After successive antibody incubation with 1:20000-diluted HRP-conjugated goat anti-mouse IgG, IgG1, or IgG2a (Abcam, UK) and washing with PBST, the assay was then developed for 6 min with 100 μl of TMB substrate solution (Solarbio, China), stopped by the addition of 50 μl of stop solution (Solarbio, China), and measured at 450 nm/630 nm (SPECTRA MAX 190, Molecular Device, USA). The endpoint titer was defined as the highest reciprocal serum dilution that yielded an absorbance > 2-fold over the optical absorbance value of the negative control.

A cytometric bead array (CBA) mouse Th1/Th2/Th17 cytokine kit (BD Biosciences) was used for cytokines detection according to the manufacturer’s protocols. Briefly, a total of 2×10⁵ splenocytes were seeded in 96-well plates and treated with purified Tul4 protein for 72 h at 37°C with 5% CO₂. Then the supernatant was collected and mixed with Capture Beads and the indicated cytokines were then quantified by flow cytometry.

A mouse IFN-γ enzyme-linked immunosorbent spot (ELISpot) kit (MabTech, Sweden) was used for T-cell epitope identification according to the manufacturer’s protocols. Briefly, a total of 2×10⁵ splenocytes from immunized mice were stimulated with 31 synthesized peptides (GL Biochem, Shanghai, 10 μg/ml) and seeded in precoated ELISpot plates for 48 h at 37°C with 5% CO₂. After successive antibody incubation and washing, the plates were measured on the AT-Spot 2100 reader (Beijing Antai Yongxin Medical Technology, China).

All of the statistical analyses were performed using GraphPad Prism 8.0.2 software. Comparisons among the groups were analyzed by one-way analysis of variance (ANOVA) unless otherwise specified. For survival analysis, the log-rank (Mantel−Cox) test was used. Antibody titer data were log transformed before analysis. Data are shown as the geometric mean with geometric SD. n.s., not significant, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.