Role of ADP-Ribosyltransferase Activity of Pertussis Toxin in Toxin-Adhesin Redundancy with Filamentous Hemagglutinin during Bordetella pertussis Infection

B. pertussis strains and growth conditions.

The B. pertussis strains used in this study are listed in Table 1. They all derived from the Tohama I strain (24) and were grown on Bordet-Gengou (BG) agar (Difco, Detroit, Mich.) supplemented with 1% glycerol, 20% defibrinated sheep blood, and 100 μg of streptomycin (Sigma Chemical Co., St Louis, Mo.)/ml, and, when appropriate, 20 μg of ampicillin (Sigma)/ml, at 37°C for 72 h. Liquid cultures of B. pertussis were performed as described previously (23) in Stainer-Scholte medium containing 1 g of heptakis(2,6-di-o-methyl) β-cyclodextrin (Sigma)/liter. For cell adherence and macrophage invasion assays, exponentially growing B. pertussis was inoculated at an optical density of 0.15 at 600 nm in 2.5 ml of Stainer-Scholte medium and grown for 24 h at 37°C. For adherence studies, the bacteria were grown in the presence of 65 μCi of l-[³⁵S]methionine plus l-[³⁵S]cysteine (NEN, Boston, Mass.)/ml. The bacteria were then washed three times in phosphate-buffered saline (PBS) and resuspended in RPMI 1640 (Gibco, Grand Island, N.Y.) at the desired density.

B. pertussis mutant constructs.

B. pertussis BPDR was obtained by double homologous recombination as described by Stibitz (36) using pGR5 (19) to delete the fhaB gene from the chromosome of PT-deficient B. pertussis BPRA (2). Consequently, this strain is PT and FHA deficient. B. pertussis BPDR-RE was constructed by introducing pPT-RE into the chromosome of BPDR by a single homologous recombination event. pPT-RE is a pPT2 derivative (1), containing the PT genes with mutations resulting in the R9K and E129G substitutions in the S1 subunit (15). The recombinant strain was selected on BG plates containing streptomycin and ampicillin. All strains produced normal amounts of fimbriae, as did the FHA-deficient strain BPGR4 (19).

Cells and growth conditions.

Human pulmonary (A549; ATCC no. CCL-185) and tracheal (HEp-2; ATCC no. CCL-23) epithelial cell lines were cultured in RPMI containing sodium penicillin G (1,000 U/ml) and streptomycin (50 μg/ml) (Gibco), 2 mM l-glutamine (Gibco), and 10% heat-inactivated fetal calf serum (Gibco) using uncoated tissue culture flasks and 24-well plates (Costar). The murine alveolar macrophage cell line MH-S (ATCC no. CRL-2019) was propagated in uncoated tissue culture flasks and 24-well plates (TPP) in the same RPMI-based medium as described above, supplemented with 1.5 g of sodium bicarbonate (Gibco)/liter, 4.5 g of glucose (Gibco)/liter, 10 mM HEPES (Gibco), 1 mM sodium pyruvate (Gibco), and 50 μM 2-β-mercaptoethanol (Gibco). Cells were detached mechanically by scraping. Dendritic cells were prepared from mouse spleen as described previously (21). Briefly, the mice received 10 μg of soluble flt3 ligand (Immunex, Seattle, Wash.) for 9 consecutive days, after which the spleen and mesenteric lymph nodes were harvested, treated with collagenase D (Boehringer Mannheim, Mannheim, Germany), and incubated with magnetic microbeads conjugated to hamster anti-mouse CD11c monoclonal antibodies (clone N418; Miltenyi Biotech, Auburn, Calif.). The CD11c⁺ dendritic cells were then purified over a magnetically activated cell sorter.

Cell adherence assay.

For 2 days in 24-well plates, 2 × 10⁵ cells per well were cultured. Cells were washed once with RPMI medium before addition of 4 × 10^{6 35}S-labeled bacteria per well and incubation for 1 h 30 min at 37°C with 5% CO₂. All bacterial strains were grown in parallel under the same conditions and harvested at the same growth phase. After three washes with RPMI to remove nonadherent bacteria, the mammalian cells were lysed with 0.5% sodium dodecyl sulfate. Whole lysates were quantified by liquid scintillation counting. The experiments were performed three times independently in triplicate.

Macrophage invasion assay.

Macrophages were grown and treated as described above. B. pertussis organisms were added at a multiplicity of infection of 20 and incubated for 1 h at 37°C with 5% CO₂. After three washes in RPMI to remove nonadherent bacteria, RPMI containing 100 μg of gentamicin (Sigma)/ml was added to kill extracellular bacteria. After 2 h, macrophages were washed three times and lysed in 300 μl of sterile distilled water. Serial dilutions were then plated onto BG agar to determine the number of viable bacteria in each well. The experiments were performed twice independently in triplicate.

i.n. infection of mice.

B. pertussis grown on BG agar was suspended in sterile PBS and adjusted to a concentration of approximately 2.5 × 10⁸ CFU/ml. Infections were performed by the intranasal (i.n.) route with 20 μl of the bacterial suspension deposited in the nostrils of 4-week-old female OF1 outbred mice (Iffa Credo, L'Arbresle, France), anesthetized (5) with a cocktail of physiological water containing 20% (vol/vol) Immalgen 1000 (Merial, Lyon, France), 10% atropine (Aguettant, Lyon, France), and 6% valium (Roche, Neuilly-sur-Seine, France), given intraperitoneally (150 μl per 20 g of body weight). At indicated time points, the lungs were aseptically removed and homogenized in PBS. Serial dilutions from individual lungs were plated onto BG agar, and the numbers of CFU were determined after 3 to 4 days of incubation at 37°C. The in vivo stability of pPT-RE in B. pertussis BPDR-RE was controlled at each time point by plating bacteria simultaneously onto BG agar containing streptomycin and onto BG agar containing streptomycin with ampicillin. The parent strain did not grow on 20 μg of ampicillin/ml after 5 days, whereas the recombinant strain formed distinct colonies after 3 days. Four OF1 mice per time point and per group of mice were assessed, and the experiments were performed five times independently. All animal studies were carried out under the guidelines of the Institut Pasteur animal study board.

TNF-α detection.

Bronchoalveolar lavage (BAL) fluids were recovered by a single lavage with 0.5 ml of sterile PBS and centrifuged for 30 s at 9,300 × g to remove cells. The BAL fluids were stored at −80°C until assayed for the presence of tumor necrosis factor alpha (TNF-α). TNF-α was quantified using the Quantikine M murine detection kit (R&D Systems, Wiesbaden, Germany). Results are expressed as picograms per milliliter. Experiments were done three times independently in triplicate.

Statistical analysis.

The results were analyzed using the unpaired Student t test. Differences were considered significant at a P value of <0.05.

Redundancy of FHA and PT in lung colonization by B. pertussis.

In order to determine the relative contributions of FHA and PT in lung colonization, mice were i.n. infected with B. pertussis BPSM or with its derivatives BPRA (ΔPT), BPGR4 (ΔFHA), or BPDR (ΔFHA ΔPT). BPSM showed a typical colonization profile with a multiplication peak 3 days after inoculation followed by a progressive clearance during the next 3 weeks (Fig. 1A). The colonization profiles obtained with BPRA (Fig. 1A) and BPGR4 (Fig. 1B) were similar to that of BPSM, indicating that the absence of either PT or FHA did not significantly affect the lung colonization.

When both PT and FHA were absent, such as in B. pertussis BPDR, no multiplication peak was observed (Fig. 1B), indicating that either PT or FHA is needed for optimal colonization of B. pertussis in mice.

In vitro adherence of B. pertussis strains.

As PT has been proposed elsewhere to act as an adhesin (40, 41), we studied the adherence of BPSM and its derivatives BPRA, BPGR4, and BPDR to the following cell lines: the murine alveolar macrophages MH-S, the human tracheal epithelial cells HEp-2, and the human pulmonary epithelial cells A549. We also tested adherence to a primary culture of dendritic cells differentiated from murine splenocytes. The absence of PT did not affect adherence to either cell line (Fig. 2), whereas adherence of the FHA-deficient BPGR4 and BPDR strains, regardless of the presence of PT, was significantly reduced compared to that of the parent strain. Moreover, the addition of up to 25 μg of purified PT/ml to BPRA did not increase adherence to the MH-S cells (data not shown).

To test whether PT could affect the invasiveness of B. pertussis into macrophages, we compared the invasion by PT-deficient strains with that by PT-producing isogenic strains. As shown in Fig. 3, the PT-deficient BPRA invaded MH-S cells as efficiently as did BPSM. On the other hand, the FHA-deficient BPGR4 and BPDR strains were not internalized.

These results indicate that FHA, but not PT, is involved in adherence to pulmonary and tracheal epithelial cell lines and in invasion of B. pertussis into macrophages.

Role of the enzymatic activity of PT in the lung colonization by FHA-deficient B. pertussis.

Since PT did not appear to act as an adhesin in the in vitro tests, we investigated the role of its ADP-ribosyltransferase activity in the redundancy with FHA in establishing colonization in vivo. An FHA-deficient B. pertussis strain producing an enzymatically inactive PT was therefore constructed. The ptx genes harboring mutations leading to the R9K and E129G substitutions in the S1 subunit of PT were introduced by homologous recombination into the chromosome of BPDR, resulting in the BPDR-RE strain. Both substitutions abolish the ADP-ribosyltransferase activity of PT (15). Immunoblot analysis of culture supernatants showed that PT-RE is efficiently produced and secreted by BPDR-RE (data not shown). The genetic constructs were stable even in the absence of selective pressure. No growth deficit was observed compared to the wild-type strain or the toxin-deletion mutant. As shown in Fig. 4, following an i.n. inoculation of BPDR-RE, no significant increase in the numbers of bacteria was observed in the lungs between days 3 and 5 compared to 3 h postinfection, similar to what was observed with the FHA- and PT-deficient BPDR strain. These results demonstrate that production of enzymatically active PT is important for cooperativity with FHA during B. pertussis infection of the mouse respiratory tract. Moreover, no necrotic lesions were observed in mouse lungs during the course of infection with BPDR-RE, in contrast to what is seen after infection with active toxin-producing B. pertussis strains, indicating that the ADP-ribosyltransferase activity of PT is responsible for the lung lesions observed after an i.n. infection with B. pertussis, consistent with earlier reports (12). Interestingly, we observed a rapid decrease of the number of BPDR-RE organisms beginning 10 days after the challenge, with a total clearance in less than 20 days. This was not seen in mice infected with BPDR, suggesting that the production of inactive PT reduces the persistence of B. pertussis in mouse lungs more than does the absence of PT. The bacteria isolated from the animals over the entire duration of the experiment remained resistant to ampicillin and produced PT-RE, indicating that the construct was stable in vivo.

Local production of TNF-α.

Since the enzymatic ADP-ribosyltransferase activity of PT appears to be important for the observed redundancy with FHA in infection, we reasoned that this might be due to the pharmacological effects of PT on the host immune system rather than to its putative role as an adhesin. This prompted us to investigate the production of TNF-α, a cytokine transiently released into the BAL fluids early during B. pertussis infection (29) that induces inflammation which might facilitate infection even of FHA-deficient strains. Alternatively, it might up-regulate the immune response and facilitate clearance.

As shown in Fig. 5, after an initial burst of TNF-α in the BAL fluids of mice infected with any one of the three B. pertussis strains, only mice infected with BPGR4, producing active PT, showed a transiently increased level of TNF-α in BAL fluids 6 days after infection. In the mice infected with either BPDR or BPDR-RE producing inactive PT, the levels of TNF-α in the BAL fluids remained close to the baseline. The levels of TNF-α induced after infection with BPGR4 were similar to those observed previously with other PT-producing strains (29). These results indicate that the production of enzymatically active PT induces the release of TNF-α in the BAL fluids upon i.n. infection with B. pertussis and suggest that this may be the basis for the redundancy with FHA to optimize B. pertussis infection.

Coinfection with BPDR and BPDR-RE.

Since the production of inactive toxin by B. pertussis results in faster clearing from the lungs than that in the absence of toxin, we wondered whether even enzymatically inactive toxin might induce general immunity against B. pertussis. This could conceivably occur via the binding activities of the B oligomer, since it has been shown elsewhere that the B oligomer expresses mitogenic activity on lymphocytes independently of the ADP-ribosyltransferase activity of PT (15, 25), which could of course increase the immune responses against B. pertussis. In that case, coinfection with BPDR and BPDR-RE would be expected to result in rapid clearing of both strains simultaneously. We therefore coinfected OF1 mice with both strains and monitored their presence in the lungs over time. The two strains could be distinguished because of their different antibiotic resistance patterns. As shown in Fig. 6, the course of colonization of BPDR was not affected by the presence of BPDR-RE in the lungs of the same mouse. Only BPDR-RE was rapidly cleared, and BPDR persisted longer. These results show that the production of enzymatically inactive yet mitogenic toxin does not induce general clearance but that the more rapid clearance of BPDR-RE than of BPDR is somehow due to the presence of enzymatically inactive PT on the surface of that particular B. pertussis organism.