IL-17 Is a Critical Component of Vaccine-induced Protection against Lung Infection by LPS-heterologous Strains of Pseudomonas aeruginosa

Bacterial strains

The bacterial strains and plasmids used in these experiments, along with their relevant characteristics and sources, are listed in Table I. PA14ΔaroA was constructed as previously described (6).

Preparation of bacterial inocula for in vivo challenge studies

Frozen bacterial stocks were plated and grown overnight at 37°C on tryptic soy agar (TSA) or TSA containing the appropriate antibiotics (carbenicillin at 400 μg/mL, gentamicin at 150 μg/mL). For intranasal inoculation experiments, bacteria were suspended in PBS. For some survival experiments, bacteria were suspended in PBS containing 1% heat-inactivated fetal calf serum (FCS; Hyclone, Logan, UT). Concentrations were adjusted spectrophotometrically and confirmed after serially diluting in PBS containing 1% FCS and enumerating growth on TSA after overnight incubation at 37°C.

Immunization of mice

Six- to 8-week-old female C3H/HeN mice were purchased from Harlan Sprague-Dawley Farms (Chicago, IL) while C57BL/6 mice were purchased from Taconic (Hudson, NY). The C57BL/6 mice were age- and gender-matched with IL-17R knockout (KO) mice (also in C57BL/6 background) that were originally obtained from Amgen (Thousand Oaks, CA) and bred in-house. Mice were housed under virus-free conditions, and all animal experiments complied with institutional and federal guidelines regarding the use of animals in research. Immunizations were performed as described previously (6). Briefly, mice anesthetized with ketamine and xylazine were inoculated by placing 10 μL of the live P. aeruginosa ΔaroA vaccine strain or live E. coli HB101 as a control on each nostril. Escalating doses of 1 × 10⁸ CFU, 5 × 10⁸ CFU, and 1 × 10⁹ CFU were administered at weekly intervals. For passive immunization experiments, mice were given pooled mouse antiserum (0.4 mL i.p.) 1 day prior to challenge. IL-17 depletion studies were done using IgG (1 mg i.p.) purified with protein G-linked agarose (Invitrogen, Carlsbad, CA) from an anti-mouse IL-17 rabbit antiserum kindly provided by A. Tzianabos (9). Control rabbit IgG was purchased from Sigma (St. Louis, MO).

Murine pneumonia model

All animal experiments complied with institutional and federal guidelines regarding the use of animals in research. We used our previously described model of acute fatal pneumonia following intranasal inoculation (10). In this model, 67-100% of the bacterial inoculum is consistently and rapidly translocated to the lungs from the nose of anesthetized 6- to 8-week-old mice (10) . For quantitation of CFU, neutrophils, and cytokines in bronchoalveolar lavage fluid (BALF), mice were sacrificed using CO₂ at the indicated time points and then BALF obtained by cannulation of the trachea followed by two instillations of 1 mL of cold PBS containing 0.5 mM EDTA. For CFU measurements, BALF was diluted in 1% proteose peptone (BD, Franklin Lakes, NJ) and then plated on TSA for enumeration of CFU after overnight growth at 37°C. The limit of detection was 1 CFU in 100 μL of the undiluted BALF, which corresponded to 10 CFU per mL. BALF was centrifuged and the supernatant frozen at −80 °C prior to measurement of IL-17 by ELISA (R&D Systems, Minneapolis, MN). Erythrocytes in the cell pellets were lysed using a Mouse Erythrocyte Lyse Kit (R&D Systems) according to the manufacturer's instructions. Total cell counts were determined with a hemacytometer and Trypan Blue counterstaining. Percent neutrophils was measured by flow cytometry using PE-labeled anti-Ly6G (1A8) after treatment with anti-CD16/CD32 (2.4G2, FcBlock, BD Biosciences, San Jose, CA). For survival analyses, mice were followed daily for at least 7 days to assess mortality. Moribund animals were sacrificed and considered nonsurvivors.

Opsonophagocytic assays

Standard methods were employed, as described (6), with infant rabbit serum (Accurate Chemical, Westbury NY) as the complement source and polymorphonuclear leukocytes (PMNs) isolated from human volunteers. By convention, the serum concentrations listed were the input rather than final concentration (the initial serum dilution is further diluted 4-fold in the assay tube). Under routine conditions, killing of ≥50% is considered biologically significant and therefore serves to classify a serum as positive for opsonic killing activity. Although killing <50% is sometimes statistically significant, this level of killing is not considered biologically significant.

Anti-P. aeruginosa IgG titers

P. aeruginosa-specific IgG binding curves were measured by ELISAs using plates coated with whole live bacteria as previously described (6).

T cell proliferation, stimulation, and cytokine measurements

Splenic T cells were isolated using CD3 cell selection columns from R&D Systems. Cell purity was routinely >90% CD3⁺ cells as assessed by flow cytometry (not shown). Lung leukocytes were isolated after perfusion of the lungs in situ with PBS containing heparin (10 Units/mL) followed by incubation with EDTA, then digestion with collagenase and DNase, then sedimentation over a discontinuous Percoll gradient, all as previously reported (11). Bacteria used as antigens were killed by incubation of a 10⁹ CFU/mL suspension with gentamicin (1 mg/mL in PBS) for 1 h at 37 °C, with killing verified by absence of growth of 100 μL samples plated on TSA. Staphylococcal enterotoxin A (SEA) was purchased from Sigma (St. Louis, MO) and used at final concentration of 10 ng/mL. For proliferation experiments, each well of a 96-well round-bottom plate was seeded with 1×10⁵ T cells, 1×10⁵ irradiated (1500 rad) splenocytes isolated from unimmunized C3H/HeN mice and purified using density centrifugation over Lympholyte-M (Accurate Chemical), and 1×10⁶ gentamicin-killed bacteria, all suspended in RPMI containing 10% heat-inactivated FCS as well as standard concentrations of glutamine, sodium pyruvate, nonessential amino acids, penicillin, streptomycin, and 1 mM 2-mercaptoethanol (all from Invitrogen, Carlsbad, CA). On day 5 of stimulation, wells were pulsed with [³H]thymidine (1 μCi/well) for 8 hours prior to scintillation counting.

For cytokine secretion experiments, 96-well plates were used containing 1×10⁵ T cells and 1×10⁵ irradiated splenocytes as feeder cells. For both proliferation and cytokine secretion experiments, anti-CD4 (GK1.5), anti-CD8 (53-6.7), and rat IgG isotype controls were purchased from BD Biosciences and used at 1 μg per well. For stimulation of bulk lung leukocytes isolated 18 h after challenge with live bacteria, no feeder cells were added. IL-17 and IFN-γ were measured in supernatants after 7 days of stimulation using ELISA (R&D Systems). Intracellular cytokine staining was performed using a kit from BD Biosciences according to the manufacturers instructions. Lung leukocytes isolated as described above 2 h after challenge with live bacteria were stimulated for 15 h ex vivo with PMA and ionomycin (Leukocyte Activation Cocktail, BD Biosciences) in the presence of Brefeldin A (GolgiPlug, BD Biosciences). After treatment with anti-CD16/CD32, cells were stained with FITC-labeled antibodies to mouse TCRβ (H57-597) or CD3ε (145-2C11), followed by fixation and permeabilization and then staining with PE-labeled anti-mouse IL-17 (TC11-18H10.1) and analysis on a MoFlo (Dako, Carpinteria, CA) flow cytometer. Antibodies and appropriate isotype controls were obtained from BD Biosciences.

Statistical analyses

Nonparametric data were evaluated by Mann Whitney U test (for 2-group comparisons) or Kruskal Wallis (for 3 or more groups) with Dunn's multiple comparison test for pairwise comparisons. Parametric data were analyzed by t test (for 2-group comparisons) or ANOVA with Dunnett's multiple comparison test using E. coli-immunized mice or cells as the control comparator group. All analyses were performed using Prism software (GraphPad Software, San Diego, CA). Survival data were analyzed by Fisher's exact test or by survival analysis with the Kaplan-Meier method, also using Prism.

Protective efficacy of PA14ΔaroA vaccination

We initially evaluated the protective efficacy of the PA14ΔaroA vaccine against LPS-homologous and heterologous strains. In our prior work, we showed that the serogroup O2/O5 vaccine PAO1ΔaroA elicited high levels of opsonic antibody as well as protection from lethality after lung challenge with LPS-homologous strains but not LPS-heterologous strains. As shown in Table II, nasal immunization with PA14ΔaroA resulted in significantly better survival compared with E. coli-immunized mice after challenge with a high dose of the parental strain PA14. There was no protection after challenge of PAO1ΔaroA-immunized mice with a lower dose of strain PA14. Notably, PA14ΔaroA-immunized mice were protected from lethality after challenge with ExoU-producing strains of the LPS-heterologous strains PAO1 (serogroup O2/O5) and PAO6a,d (serogroup O6). There was no protection, however, after challenge with the non-cytotoxic strains Fisher IT-7 (serogroup O2/O5) or 6294 (serogroup O6). This is likely related to the higher LD₅₀s of the non-cytotoxic strains in this model and thus the very high challenge doses required to achieve 100% mortality in the E. coli-immunized control mice.

Opsonic killing activity in antisera raised to P. aeruginosa PA14ΔaroA

For most vaccines effective against P. aeruginosa infections, serum opsonic antibody directed against the LPS O antigen has been the most important immune effector (12). To assess the role of serum antibodies and antibodies to specific antigens such as the ExoU cytotoxin in the protection afforded by the PA14ΔaroA vaccine, we first tested the opsonic killing activity of pooled sera obtained from immunized mice 3 weeks after the third (final) nasal immunization. As depicted in Fig. 1A, there was minimal opsonic killing of strain PAO1 in serum from mice immunized with the PA14ΔaroA strain, whereas good killing was achieved using sera from mice immunized with the LPS-homologous PAO1ΔaroA strain. Opsonic killing activity in antisera raised to PA14ΔaroA was present against the parental PA14 strain, although the opsonic titer (the calculated serum dilution at which the killing activity dropped to 50% as determined by linear regression) was approximately 50 (not shown), which is much lower than the titer of approximately 1000 seen with antiserum to PAO1ΔaroA against its parental strain PAO1 (7). Because PA14ΔaroA possesses the genes for ExoU while PAO1ΔaroA does not, and thus the PA14-based vaccine could induce antibody to this antigen, we next evaluated whether antibody to ExoU could be mediating opsonic killing if for unexpected reasons this antigen was retained on the bacterial cell surface. As shown in Fig 1B, there was no significant opsonic killing of an ExoU-producing strain compared with the strain containing the empty vector.

Passive immunization and whole-cell ELISA experiments

Next, we determined whether passive immunization with pooled antisera from PA14ΔaroA-immunized mice mediated protection against lethal pneumonia due to the LPS-heterologous strain ExoU⁺ PAO1. Mice given the PA14ΔaroA antiserum i.p. 1 day prior to challenge had significantly higher mortality compared with the 0% mortality of mice given identical amounts of antisera raised to the LPS-homologous PAO1ΔaroA strain (Fig. 1C). These results suggested, surprisingly, that an antiserum-independent mechanism is a key component of the heterologous protection engendered by active vaccination with the PA14ΔaroA vaccine. We also tested for O antigen-specific and ExoU-specific antibodies by comparing IgG titers of the PAO1ΔaroA- and PA14ΔaroA-immunized mouse antisera to whole cells of the wild-type parental strain PAO1 (containing the empty vector) compared with strain ExoU⁺ PAO1 as well as ExoU-producing and non-producing LPS-rough galU mutants of strain PAO1 (galU mutants are O antigen deficient with a truncated outer LPS core (13)). As shown in Fig. 2, the PA14ΔaroA antisera had the expected lower reactivity than did the antisera raised to PAO1ΔaroA against P. aeruginosa strains PAO1 (pUCP19) and ExoU⁺ PAO1 based on its different LPS serogroup. The titers to the O antigen-deficient galU mutants were identical, suggesting that the PA14ΔaroA vaccine elicited a comparable level of antibody to non-O-side-chain antigens as did the PAO1ΔaroA vaccine.

T cell responses elicited by P. aeruginosa PA14ΔaroA vaccine

To determine whether immune T cells are induced by nasal immunization with the P. aeruginosa ΔaroA vaccines, we isolated splenic T cells 3 weeks after immunization (prior to challenge) and co-cultured them with irradiated splenocytes as antigen-presenting cells (APCs) and gentamicin-killed P. aeruginosa as antigen and measured T cell proliferation by [³H]thymidine incorporation after 5 days. As shown in Fig. 3A, there was a high level of proliferation of both PAO1ΔaroA and PA14ΔaroA-immune T cells to either PAO1 or PA14, although the PA14ΔaroA-immune T cells proliferated less well when stimulated with PAO1 compared with the PAO1ΔaroA-immune T cells. Proliferation of PAO1ΔaroA-immune T cells was inhibited by an antibody to CD4 but not to CD8 (Fig. 3B). Proliferation of PAO1ΔaroA and PA14ΔaroA-immune T cells was high when stimulated with multiple P. aeruginosa strains regardless of LPS serogroup or ability to produce ExoU (Figs. 3C). Interestingly, the P. aeruginosa ΔaroA-immune T cells proliferated when cultured with E. coli but the E. coli-immune T cells were not significantly stimulated by P. aeruginosa strains.

Cytokine secretion by P. aeruginosa ΔaroA vaccine-elicited T cells

As neutrophils are known to be essential mediators of protective immunity in the lung against P. aeruginosa infection (14, 15), we hypothesized that the recently described Th17 subset of CD4 helper T cells, which secrete the neutrophil-attracting cytokine IL-17, might play a role in rapid recruitment of neutrophils to the airways. We thus evaluated IL-17 secretion after 7-day co-culture of splenic T cells from naïve and immunized mice cultured with irradiated splenocytes and gentamicin-killed P. aeruginosa strains PAO1 and PA14 as antigens (Figs. 4A and 4B). For comparison, we also measured levels of IFN-γ, the predominant Th1 effector cytokine, and found them not to be significantly different among the vaccine groups. In contrast, the levels of IL-17 in the supernatants of the P. aeruginosa ΔaroA-immune T cells stimulated with P. aeruginosa were significantly higher than those of control T cells. Addition of the anti-CD4 monoclonal antibody (clone GK1.5, which blocks class II MHC antigen-specific binding to CD4 and thereby blocks CD4 T cell function (16)) during co-culture returned the IL-17 levels to those of the control T cells (Fig. 4B), indicating that CD4 T cells are the predominant source of IL-17 in this system. We also isolated intraparenchymal and airway lung leukocytes 18 h after challenge of immune and non-immune mice by digesting the lung tissue with collagenase and DNase followed by a discontinuous Percoll gradient. Numbers of T cells in the lungs prior to challenge were too low to allow accurate assessment of cytokine secretion from cells prior to challenge so we therefore only evaluated cytokine secretion from lung T cells after challenge. At the 18-h time point after challenge, the cellular infiltrate is approximately 35% lymphocytes (with the remainder neutrophils) as assessed by forward-and side-scatter FACS analysis; and the composition was not significantly different between the experimental groups (not shown). The isolated lung leukocytes were co-cultured with gentamicin-killed PAO1 for 7 days. The IL-17 levels in the supernatants of these in vivo-immunized, ex vivo-stimulated lung leukocytes are shown in Fig. 4C and were remarkably higher in the ΔaroA-immune groups, with the PAO1ΔaroA-immune and PA14ΔaroA-immune T cells having approximately 10-fold and 40-fold higher IL-17, respectively, than the control T cell groups.

To determine the cellular source of the IL-17 in vivo, we measured intracellular staining for IL-17 in lung parenchymal and airway lymphocytes isolated as described above but from mice sacrificed just 2 h after challenge with live bacteria. Again, numbers of lymphocytes in the lungs of immunized mice prior to challenge were too low to allow intracellular cytokine analysis. Cells were stimulated for 15 h with PMA and ionomycin in the presence of brefeldin A. As shown in Fig. 4D, there was a small population of IL-17-producing TCRβ-positive cells in the lungs of ΔaroA-immune mice, with 3-fold more in the PA14ΔaroA-immunized group. Similar proportions of IL-17-positive cells were seen in CD3ε-positive cells, and no IL-17-positive cells were seen outside the lymphocyte gate of ΔaroA-immune mice (not shown). We were unable to assess CD4-staining in these experiments due to apparent decreased expression of CD4 during stimulation (not shown). Unstimulated cells showed no detectable IL-17 (not shown). Lung leukocytes (lymphocytes and non-lymphocytes) isolated 18 h after challenge with live bacteria showed no intracellular staining for IL-17 with or without PMA/ionomycin stimulation (not shown).

Neutrophil recruitment to the lungs of P. aeruginosa ΔaroA-immunized mice

Because IL-17 is known to stimulate neutrophil recruitment to the lung, we tested whether the numbers of neutrophils recruited to the airways are greater in the P. aeruginosa ΔaroA-immunized mice and whether the different P. aeruginosa ΔaroA vaccines elicited different neutrophil recruitment. As shown in Fig. 5A, there were significantly more neutrophils in BALF 6 h after challenge of P. aeruginosa ΔaroA-immunized mice compared with the E. coli-immunized and unimmunized controls. Notably, even at this early time point after challenge, there were significantly fewer CFU of P. aeruginosa in the BALF (Fig. 5B) and homogenized lung tissue (not shown) in the P. aeruginosa ΔaroA-immunized mice compared with the control groups. IL-17 levels in BALF (Fig. 5C) obtained 6 h after challenge with a lethal dose of ExoU⁺ PAO1 were significantly higher in the PA14ΔaroA-immunized compared with the E. coli-immunized mice (p <0.01, ANOVA with Dunnett's multiple comparison test). At the later time point (18 h), neutrophil numbers and IL-17 levels in BALF were highest in the E. coli-immunized mice. Bacterial CFU were very high in the control groups at this time after challenge, indicating that these neutrophils were not effective at clearing the challenge strain.

Effect of loss of IL-17 on protective efficacy of P. aeruginosa ΔaroA vaccines against LPS-heterologous strain challenge

We next determined the effects of neutralization of IL-17 prior to lung challenge with the LPS-heterologous strain PAO1 ExoU⁺ in mice immunized intranasally with the P. aeruginosa PA14ΔaroA vaccine or with E. coli as control. As expected, there was no effect of neutralization of IL-17 on survival following challenge of the E. coli-immunized mice (left panel of Fig. 6A). However, there was significantly higher mortality (p <0.02, log rank test) in the PA14ΔaroA-immunized mice that received anti-IL-17 IgG compared with the mice given control IgG (Fig. 6A, right panel), indicating that depletion of IL-17 abrogates vaccine-induced protection. In contrast, mice immunized with the PAO1ΔaroA vaccine and then similarly depleted of IL-17 had no change in median survival and were fully protected against LPS-homologous strain PAO1 ExoU⁺ because these mice had high levels of serum opsonic antibody specific for the LPS of strain PAO1 (not shown). We also found that when we immunized mice genetically deficient in production of the IL-17 receptor (IL-17R KO) with PA14ΔaroA (or with E. coli as control) and then challenged them with ExoU⁺ PAO1, the IL-17R KO mice had significantly decreased survival compared with wild-type mice (Fig. 6B, right panel). We observed lower percent survival in the PA14ΔaroA-immunized wild-type C57BL/6 mice challenged with 5×10⁷ CFU/mouse (40% survival, Fig. 6B) compared with PA14ΔaroA-immunized wild-type C3H/HeN mice given the same challenge dose (88%, Table 2). This was likely due to the high susceptibility of mice in the C57BL/6 background to P. aeruginosa lung infection (17) and the relative resistance of C3H/HeN mice (18).

Consistent with the IL-17 neutralization experiments in PAO1ΔaroA-immunized C3H/HeN mice, IL-17R KO mice immunized with the PAO1ΔaroA vaccine and then challenged with the LPS-homologous strain ExoU⁺ PAO1 were fully protected (not shown). These PAO1ΔaroA-immunized IL-17R KO mice had normal levels of opsonic antibody to strain PAO1 (not shown), which very likely accounted for the lack of an impact of deficient IL-17 signaling. We next evaluated the neutrophil recruitment to the airways 6 h after challenge of PA14ΔaroA-immunized wild-type C57BL/6 mice compared to PA14ΔaroA-immunized IL-17R KO mice. We again used P. aeruginosa strain ExoU⁺ PAO1 (7×10⁶ CFU/mouse) as the challenge strain. As shown in Fig. 6C, there were significantly fewer BALF neutrophils normalized to bacterial load in the IL-17R KO mice. Of note, unimmunized IL-17R KO and wild-type C57BL/6 mice had similar survival curves when challenged with 2×10⁶ CFU of strain ExoU⁺ PAO1 (Fig. 6D) or 7×10⁵ CFU (100% survival in both groups, n=6 mice per group, not shown), indicating that IL-17R signaling is not important for the innate host defense against acute lethal pneumonia caused by P. aeruginosa.