The K. pneumoniae 260 strain was provided by Professor Yigang Tong and isolated from the Respiratory Department of the 307 Hospital of the Chinese People’s Liberation Army. K. pneumoniae 260 is resistant to a variety of antibiotics (Amikacin, Cefuroxime-Sodium, Ciprofloxacin, Piperacillin, etc.). K. pneumoniae 260 strain was grown at 37°C in Luria–Bertani broth. Before bacterial challenge, overnight cultures of K. pneumoniae 260 strain was sub-cultured and harvested until OD600 ~ 1. K. pneumoniae 260 strain was then collected by centrifugation, washed once, and resuspended in PBS.

For expression of recombinant proteins (11), Escherichia coli BL21 strain (DE) was grown in Luria–Bertani broth containing 40 μg/ml kanamycin up to OD600 ~ 0.5. One liter of bacterial culture was induced with 0.5–1 mM isopropyl-β-D1-thiogalactopyranoside (IPTG) and grown for 4 h at 37°C.

The workflow is shown in Figure 1A . Bacterial cells were harvested by centrifugation at 3,500 × g for 10 min, at 4°C, and washed three times with PBS, pH 7.4. Cell pellets were resuspended in PBS containing 30% sucrose (pH 7.4) as trypsin incubation buffer. Tryptic digestions were carried out with porcine sequencing grade modified trypsin (Promega) for different times at 37°C. Samples of the bacterial suspension were taken pre- and post-protease digestion to determine colony-forming units (CFUs). Bacterial cells were removed by centrifugation (3500 × g, 10 min, 4°C). The supernatants (the “surfome” containing the peptides) were filtered through 0.22-μm-pore-size filters to remove any remaining bacterial cells. “Surfome” was digested with 2 μg of trypsin overnight at 37°C with agitation. Then, the samples were desalted prior to MS and eventually eluted in 70% ACN and 0.1% formic acid. Afterwards, the sample was then placed in SpeedVac to allow the solvent to evaporate completely, and the resulting dry sample was stored at −20°C for use. The solvent was completely vaporized by SPEEDVAC, and the resulting dry sample was stored at −20°C for mass spectrometric (MS) anylsis. After mass spectrometry (MS) conducted by MS machine (Orbitrap Fusion™ Tribrid™ Mass Spectrometer, thermo), the database of the K. pneumoniae strains from UniProt was used. The identification of the MS results was performed with Thermo Proteome Discoverer (version 1.4.1.14).

Complete genomic sequences of 389 K. pneumoniae strains were downloaded from GenBank (ftp://ftp.ncbi.nlm.nih.gov/genomes/, accessed: 11/18/2019). The potential KOMP sequences were then aligned to genomic sequences with the help of blast from BLAST (12); all 389 genomes obtained at least one protein sequences with at least 60% identity and 70% coverage. Multiple sequence alignment (MSA) was executed by MUSCLE with default parameter using the sequences obtained above (13). The Valdar scoring method was applied for the calculation of residues conservation score as it incorporates sequence redundancy and alignment gaps; calculation was performed on Jalview Version 2 using online method AAcon (Amino Acid Conservation) (14, 15).

DNA sequences of KOMPs were amplified by PCR using genomic DNA of K. pneumoniae 260 strain as template. The PCR products were then cloned into pET28a vector. Next, the recombinant plasmids were transformed into E. coli BL21 strain (DE). A single transformed BL21 clone was subjected to induction with IPTG followed by testing for protein expression by SDS-PAGE analysis. Bacteria were harvested by centrifugation and resuspend in binding buffer (Tris–HCl, pH 8.0, 20 mM imidazole) followed by sonication. After centrifugation at 15,000 × g for 30 min at 4°C, the soluble cell extracts were collected and filtered using a 0.22-μm membrane. The samples were then loaded on a nickel-activated chelating Sepharose column (GE Healthcare). After washing, the bound proteins were eluted with elution buffer (Tris–HCl, pH 8.0, 250 mM imidazole). The expression of purified proteins was identified using SDS-PAGE analysis with anti-histidine antibodies. The proteins were subject to endotoxin removal afterwards (Pierce).

Five KOMPs formulated with aluminum hydroxide gel adjuvant (Alum, 1 mg/ml) was injected subcutaneously into BALB/c mice (10 per group), respectively. The mice were subjected to booster vaccinations every 2 weeks twice. Alum adjuvant with 25 μg of each purified protein was used to immunize animals. Mice of mock group received an equal amount of Alum adjuvant in PBS. The antibody titers to each antigen in serum samples collected from animals were documented by enzyme-linked immunosorbent assay (ELISA) after each booster vaccination as described elsewhere (16).

Active (10 days after second booster vaccination) immunized BALB/C mice were challenged with a lethal dose of 1 × 10⁹ CFU K. pneumoniae 260 by tail vein intravenous injection. The total volume of the injection was 100 μl. The wellbeing of infected mice was monitored daily for 14 days.

K. pneumoniae 260 was used in the pneumonia model. After anesthesia with ketamine (100 mg/kg) and xylazine (5 mg/kg), BALB/c mice were challenged intranasally with 1 × 10⁹ CFU of K. pneumoniae 260. The total volume of the inoculation was 50 μl. The wellbeing of infected mice was monitored daily for 144 h.

Seven days after the third booster vaccination, animals were challenged with a sub-lethal dose, 5 × 10⁸ CFU, of K. pneumoniae 260. The lungs, kidneys, and spleens were collected 4 days after the sub-lethal challenge and homogenized. CFUs were enumerated following serial diluting and plating on BHI agar.

Mice were sacrificed 7 days after the second booster vaccination. After euthanasia, spleens were collected and single suspensions of splenocytes were obtained (11). Interferon gamma (IFN-γ)-, interleukin 4 (IL-4)-, and interleukin 17A (IL-17A)-producing splenocytes from vaccinated or control mice were analyzed using a cytokine-specific enzyme-linked Immunospot assay (ELISPOT, R&D Systems, United States) as described by the manufacturer. Briefly, splenocytes isolated from immunized mice were plated at a concentration of 1 × 10⁵ cells/well and induced with each antigen alone (0.2 μg/well) or combined (0.1 μg of each antigen/well) in triplicate and incubated for 20 h at 37°C. Ionomycin (1 μg/ml, Sigma, USA) and phorbol myristate acetate (PMA, 50 ng/ml, Sigma) were used as positive controls. Splenocytes from unstimulated, immunized mice and RPMI 1640-treated splenocytes were used as negative controls. After the cells were decanted, biotinylated primary monoclonal antibodies were added to each well and the plates were incubated for 1 h at 37°C. The plates were incubated with streptavidin–HRP conjugate for 1 h at 37°C and subjected to color development with TMB solution. Finally, the spots were enumerated using an Immunospot analyzer.

The opsonophagocytosis killing assay (OPKA) was performed as previously described (17). Human promyelocytic leukemia cells HL-60 were differentiated into phagocytes using 0.8% DMF. K. pneumoniae 260 grown overnight were washed once in PBS and resuspended in HBSS buffer (Ca²⁺/Mg²⁺). The bacteria were incubated with heat-activated mouse antiserum against different KOMPs at 4°C for 20 min. Differentiated HL-60 cells were distributed at 3.7 × 10⁶/well with MOI ~ 50:1 in the presence of 10% (v/v) rabbit complement. Following incubation at 37°C for 1 h with agitation at 600 rpm, samples were plated on BHI agar plates for CFU enumeration.

BALB/c mice were supplied by the Laboratory Animal Unit (LAU) of the University of Hong Kong. All mouse experiments were approved by the Committee on the Use of Live Animal in Teaching and Research of the University of Hong Kong (CULATR 4493-17).

At least two independent experiments, run under the same conditions, were performed for all studies. All data were analyzed in GraphPad prism 8.0 software (GraphPad Software Inc., CA, USA). For the blood infection and pneumonia models, statistical significance was assessed with the log-rank (Mantel-Cox) analysis. The Student’s paired t-test was used to analyze the statistical significance of OPKA experiments and bacterial load measurements.

Both the “shaving” experiments and the MS experiments were repeated three times independently. After comparing the data from all three experiments, only peptides that were detected in all three experiments were accepted. A total of 562 protein candidates were obtained. Among the 562 proteins, there were 12.36% of membrane proteins, 2.03% of periplasm proteins, 34.32% of cytoplasm proteins, and 51.29% of unknown protein ( Figure 1B ). After analyzing 389 K. pneumoniae genome sequences available in NCBI databases, we obtained five proteins that are highly conserved in all strains selected. The amino acid sequence identity ranges from 96% to 100% ( Figures 1B, C ); we name them Kpn_Omp001 (GenBank: QDX70306.1), Kpn_Omp002 (GenBank: QEX42815.1), Kpn_Omp003 (GenBank: QER79328.1), Kpn_Omp004 (GenBank: QEX41443.1), and Kpn_Omp005 (GenBank: QDX64613.1). Recombinant KOMPs were expressed in E. coli BL21 and purified using a three-step chromatography strategy. Results indicated that the majority were KOMPs expressed in the soluble form in a high yield (>90%).

To measure the antibody-mediated responses of KOMPs (Kpn_Omp001, Kpn_Omp002, Kpn_Omp003, Kpn_Omp004, and Kpn_Omp005), Balb/c mice were immunized with KOMPs individually or PBS according to Figure 2A , and KOMP-specific antibody titers were determined by ELISA. High levels of IgG, IgG1, and IgG2a were observed against target antigens in all immunized mice at days 7, 21, and 35. Antibody production after two boost immunizations was considerably higher than that with one immunization. Analysis of the IgG isotype revealed that vaccination with any of the five KOMPs induced both Th1- and Th2-associated antigen-specific IgG2a and IgG1 antibody responses ( Figure 2B ). IgG, IgG1, and IgG2a specific for five KOMPs were not detected in the serum of mice mock immunized with PBS plus AHG.

One of the criteria for selecting these KOMPs as vaccine candidates was their ability to induce protection in one or more animal models of K. pneumoniae infection. Mice were immunized with each KOMP formulated with aluminum hydroxide. After immunization, mice were challenged intravenously (i.v.) with a lethal dose of 1 × 10⁹ CFU K. pneumoniae 260 ( Figure 3A ). The survival of mice was recorded in a 14-day period. In the bacteremia model, immunization with Kpn_Omp001, Kpn_Omp002, and Kpn_Omp005 resulted in significantly increased survival rate compared with the control mice immunized with only adjuvant (Alum, mock) ( Figure 3B ). More than 80% (p = 0.0007) of Kpn_Omp001, 50% (p = 0.04) of Kpn_Omp002, and 50% (p = 0.0402) of Kpn_Omp005 immunized mice survived over 14 days after lethal challenge. The survival rate of mice immunized with Kpn_Omp003 and Kpn_Omp004 did not show any protection compared with the control mice ( Figure 3B ).

Then, we evaluated the efficacy of Kpn_Omp001, Kpn_Omp002, and Kpn_Omp005 in the pneumonia model ( Figure 4A ). The survival of immunized and control mice was evaluated for a 144-h period after nasal challenge of lethal dose of K. pneumoniae strain 260. The survival rate of mice immunized with Kpn_Omp001 (50%, p = 0.0105), Kpn_Omp002 (20%, p = 0.0261), or Kpn_Omp005 (30%, p = 0.0137) was always significantly superior to that observed in mock ( Figure 4B ).

In addition, the bacteria load recovered from different organs in mice after sub-lethal dose of intravenous challenge of K. pneumoniae strain was counted ( Figure 5A ). As shown in Figure 5B , immunization with Kpn_Omp001, Kpn_Omp002, and Kpn_Omp005 resulted in a significant reduction of bacterial load when compared with the control mice immunized with alum alone. The reduction of bacteria load was most significant in kidneys. Reduction of CFUs varied from a minimum of 1 to a maximum of 2 logs.

To evaluate T-cell responses generated by these three potential KOMPs, we measured KOMP-specific IFN-γ-, IL-4-, and IL-17A-secreting splenocytes of Balb/C mice after ex vivo re-stimulation, using ELISpot assay ( Figure 6A ). These three potential KOMPs (Kpn_Omp001, Kpn_Omp002, or Kpn_Omp005) elicited a significant increase in the number of IFN-γ-, IL-4-, and IL-17A-secreting splenocytes ( Figure 6B ). These findings suggest that immunization with Kpn_Omp001, Kpn_Omp002, or Kpn_Omp005 induces strong antigen-specific Th1, IFN-γ-producing T-cell response in addition to Th2 and Th17 responses.

Next, we performed if its antiserum could facilitate opsonophagocytosis of K. pneumoniae. The antiserum samples were incubated with K. pneumoniae 260 and differentiated HL60 cells in the presence of rabbit complement. As shown in Figure 6C , serum from alum-immunized mouse did not mediate bacterial killing. Likewise, bacterial killing was not facilitated without HL-60 cells or active rabbit complement. On the contrary, HL-60 cells killed over 40%, 50%, or 30% of K. pneumoniae with the anti-Kpn_Omp001, anti-Kpn_Omp002, or anti-Kpn_Omp005 serum. These results suggest that humoral immunity may confer the protective effect of K. pneumoniae anti-Kpn_Omp001, anti-Kpn_Omp002, or anti-Kpn_Omp005 vaccination.